Drug Polym Interactions

The utilization of drugpolymer interactions forimproving the chemical stability of hot-melt extrudedsolid dispersionsZhefei Guo, Ming Lu, Yongcheng Li, Huishi Pang, Ling Lin, Xu Liu and Chuanbin Wu

School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou, China

Keywordsamorphous solid dispersion; chemical stability;hot-melt extrusion; hydrogen bond;interaction

CorrespondenceMing Lu and Chuanbin Wu, Department ofPharmaceutics, School of PharmaceuticalSciences, Sun Yat-sen University, GuangzhouHigher Education Mega Center, Guangzhou510006, China.E-mail: [email protected];[email protected]

Received May 11, 2013Accepted August 25, 2013

doi: 10.1111/jphp.12145

Abstract

Objective Interactions between drugs and polymers were utilized to lower theprocessing temperature of hot-melt extrusion (HME), and thus minimize thethermal degradation of heat-sensitive drugs during preparation of amorphoussolid dispersions.Methods Diflunisal (DIF), which would degrade upon melting, was selected as amodel drug. Hydrogen bonds between DIF and polymeric carriers (PVP K30, PVPVA64, hydroxypropyl methylcellulose and Soluplus) were revealed by differentialscanning calorimetry and Fourier transform infrared spectroscopy. The hot-meltextruded solid dispersion was characterized by powder X-ray diffraction (XRD),scanning electron microscopy (SEM) and high-performance liquid chromatogra-phy (HPLC).Key findings The results of hot-stage polar microscopy indicated that DIF wasdissolved in molten polymers at 160C, much lower than the melting point of DIF(215C). At this temperature, amorphous solid dispersions were successfully pro-duced by HME, as confirmed by XRD and SEM. The related impurities in amor-phous solid dispersions detected by HPLC were lower than 0.3%, indicating thatthermal degradation was effectively minimized. The dissolution of DIF fromamorphous solid dispersions was significantly enhanced as compared with thepure crystalline drug.Conclusion This technique based on drugpolymer interactions to preparechemically stable amorphous solid dispersions by HME provides an attractiveopportunity for development of heat-sensitive drugs.

Introduction

The amorphous form of drug has drawn considerableattention, as theoretically it represents the most energeticsolid state, and may provide the advantages in terms ofsolubility and bioavailability.[1] Recently, amorphous soliddispersions have been successfully developed to thermody-namically or kinetically stabilize the amorphous formof drugs and yield drug products with enhancedbioavailability.[25] Among the various techniques for pre-paring solid dispersions, hot-melt extrusion (HME) hasstood out with obvious advantages due to its single-step,simple and organic solvent-free preparation process, andgained increasing popularity.[68] However, the elevated pro-cessing temperature has certainly limited the application of

HME for heat-sensitive drugs, and so far only a few numberof studies focused on this problem.[9]

In many cases, thermal degradation of drug is mainlyrelated to the cumulative exposure to heat, which is a func-tion of temperature and exposing duration.[10] Techniquessuch as the formation of cocrystal have been utilizedto minimize the thermal degradation of drug. Forexample, amorphous solid dispersion of heat-sensitivecarbamazepine was successfully prepared by HME at 160C,below the melting point of drug (190C), as carbamazepineformed in-situ cocrystal with the added nicotinamide.Carbamazepinenicotinamide cocrystal did not degradeupon melting at 160C.[11] However, most of the cocrystals

bs_bs_banner

And PharmacologyJournal of Pharmacy

Research Paper

2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 285296 285

of the heat-sensitive drugs are also heat-sensitive.[12]

Another technique is to convert the crystalline heat-sensitive drug to amorphous form by solvent method. Aftersolvent treatment, the extrusion temperature could belowered much to minimize the degradation and produce anamorphous solid dispersion.[13,14] But this method relies onhigh-toxic solvent (methylene chloride), which may induceproblems such as solvent residue, pollution and high prepa-ration cost. In addition, the solid dispersion of thermallylabile p-aminosalicylic acid and ethylcellulose was preparedby HME with the aid of pressurized carbon dioxide as atemporary plasticizer.[15]

Some previous work reported that if intermolecularinteraction existed between drug and polymer, amorphoussolid dispersion can be successfully extruded under thenormal melting point of drug, such as bicalutamidepoly(ethylene oxide) solid dispersion[16] and paracetamolaminoalkyl methacrylate copolymer(Eudragit EPO) soliddispersion.[17] Although these approaches have been utilizedto facilitate the extrusion of amorphous solid dispersionsfor thermal stable drugs, few publications reported the utili-zation of drugpolymer interaction to avoid thermal degra-dation of heat-sensitive drug during HME.

In this study, diflunisal (DIF) was selected as a heat-sensitive model drug. The carboxylic group and thehydroxyl group in DIF molecule can act as hydrogen bonddonors.[18] Four polymers were chosen to possibly formhydrogen bonds with DIF, namely PVP VA64, Soluplus,hydroxypropyl methylcellulose (HPMC) and PVP K30.Their structures are shown in Figure 1. The DIFpolymerinteractions and melting point depression were studied bydifferential scanning calorimetry (DSC), hot-stage polarmicroscopy (HSPM) and Fourier transform infrared spec-

troscopy (FTIR). Based on these preliminary studies, amor-phous DIF solid dispersions were successfully prepared byHME below the melting point of pure crystalline DIF.Physical state, chemical degradation, dissolution behaviourand physicochemical stability of the solid dispersions wereinvestigated. The aim of this study was to explore drugpolymer interaction as a convenient and effective method toprepare chemically stable amorphous solid dispersions ofheat-sensitive drugs by HME.

Materials and Methods

Materials

DIF was purchased from Juhua Group Corporation (Zhe-jiang, China). Kollidon VA64 (PVP VA64) and Soluplus(Soluplus) were friendly supplied by BASF SE (Ludwigsha-fen, Germany), HPMC (2910 grade) by Colorcon Inc.(Shanghai, China), and Plasdone K-29/32 (PVP K30) byInternational Specialty Products Inc. (Wayne, NJ, USA). Allreagents used were of analytical grade or better, and all thematerials were dried before use.

Methods

Hot-melt extrusion

Solid dispersions containing different ratios of DIF/polymer(20 : 80, 30 : 70 and 40 : 60, w/w) were prepared by HME(HAAKE MiniCTW micro compounder, Thermo ElectronGmbH, Karlsruhe, Germany). The materials were premixedbefore manually fed into the extruder. The extrusion tem-perature was set at 160C, and the screw speed was fixed at150 rpm. After cooling at ambient condition, the extrudateswere milled and passed through sieves for further analysis.

O OH

OH

H

N O

nm

O OH

CH3

CH3

HPMC

OH

x

DIFFF

N O

n

PVP K30

PVP VA64

OR

O

OR

orR=H,

OR ORHO

O

OO

ON

O

lm

Soluplus

HO

n

O

R

nRO

RO OO

OH

Figure 1 The chemical structures of diflunisal and polymers used in the study.

Zhefei Guo et al.Improve chemical stability in HME

2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 285296286

Thermal gravimetric analysis

The thermal stability of DIF was evaluated by thermalgravimetric analysis (TGA) using NETZSCH STA-409thermogravimetric analyser (NETZSCH group, Germany).Samples (510 mg) were heated in aluminum pans to 250Cat a heating rate of 10C/min, with nitrogen purge at a flowrate of 40 ml/min. Data were analysed using Proteus analy-sis software (NETZSCH Group, Selb, Germany).

Differential scanning calorimetry

Physical mixtures were prepared by geometric mixing ofpure crystalline drug with polymers. The onset of meltingwas measured by a NETZSCH STA-409 thermogravimetricanalyser at a heating rate of 2C/min, with nitrogen purge at40 ml/min for calculation of the FloryHuggins interactionparameters. The Tg of the solid dispersion was measuredwith a NETZSCH DSC 200 F3 Maia at a heating rate of5C/min, with nitrogen purge at 40 ml/min. The accuratelyweighed samples (510 mg) were placed in aluminum pansfor analysis, and an empty aluminum pan was used as a ref-erence. The temperature and enthalpy were calibrated usingindium. All analyses were carried out in triplicate.

Further, the FloryHuggins interaction parameters wereestimated from the results of melting point depression. Thevolume fractions of drug and polymers were calculated bydividing the weight fraction by the true density of theamorphous materials. The true densities of the polymerswere experimentally determined using TD-2200 heliumpycnometer (Builder, Beijing, China), while the true densityof amorphous DIF was calculated as 1.437 0.06 g/cm3

using ChemSketch software (ACD Labs, Toronto, Canada).The lattice volume of the drug was defined as equal to itsmolecular volume.[19]

Powder X-ray diffraction

The powder X-ray diffraction (XRD) study was performedat ambient temperature with a powder XRD apparatus(Rigaku D/Max 2000, Tokyo, Japan) operating at 30 mAand 30 kV, with the scanning rate of 5/min and the diffrac-tion angles (2) of 540.

Fourier transform infrared spectroscopy

Intermolecular hydrogen bonds between DIF and polymerswere determined using FTIR equipped with a KBr beamsplitter (TEN-SOR37, Bruker, Bremen, Germany). Powderysamples were mixed with dry KBr (spectroscopy grade,Sinopharm Chemical Reagent Co., Ltd, Shanghai, China)before they were compressed into slides under a pressure of10 tons for 1 min. Samples were examined using the trans-mission mode with the scan range of 4004000/cm, 64

scans per spectrum, and a resolution of 4/cm. The softwareBruker OPUS 6.5 was used for data analysis.

In-vitro dissolution study

Dissolution behaviours of the solid dispersions were studiedby using a USP paddle dissolution apparatus (ZRS-8 G dis-solution tester, Tianda Tianfa Technology Co., Ltd, Tianjin,China). The solid dispersions were milled to collect parti-cles in the range of 120160 mesh for test. Samples equiva-lent to 10 mg of DIF were accurately weighed and addedinto the dissolution medium (900 ml of distilled watermaintained at 37 0.5C) with a paddle speed of 50 rpm.At each sampling point (5, 10, 20, 30, 45, 60, 90 and120 min), 5 ml of medium sample was drawn with the samevolume of fresh medium as replacement. The aliquotsamples were centrifugalized at 15 000 rpm for 15 min, andthe clear supernatants were analysed by high-performanceliquid chromatography (HPLC). All analyses were carriedout in triplicate.

High-performance liquid chromatography

DIF assay was determined by a Shimadzu LC-20AT HPLCsystem (Shimadzu Corporation, Kyoto, Japan) equippedwith a Waters Xbridge Shield RP-18 (5 m 4.6 mm 250 mm column, Waters Corporation, Milford, MA, USA)at a wavelength of 253 nm. The mobile phase was composedof methanol and 0.4% acetic acid solution (71 : 29, v/v),and the DIF peak was eluted at approximately 14 min at30C, with an injection volume of 20 l and a flow rate of1.5 ml/min. All analyses were carried out in triplicate.

Related substances of DIF were determined by the sameequipment at the wavelength of 254 nm at 30C. Themobile phase was composed of water, methanol, acetonitrileand acetic acid (55 : 23 : 30 : 2, v/v) at a flow rate of 1.5 ml/min. The test solution contained 10 mg/ml of DIF, and thereference solution contained 50 g/ml of DIF. The injectionvolume was 5 l. All analyses were carried out in triplicate.

Hot-stage polar microscopy

HSPM studies were conducted using a polarizing opticalmicroscope (Mingmei, Co. Ltd, Guangzhou, China). Eachsample was placed between a glass slide and a cover glass,and then fixed on KER 3100-08S hot stage (Kaier Co. Ltd,Nanjing, China), heated with a heating rate of 5C/min overthe range of 100160C to facilitate the dissolving process.The morphology changes during the heating process wererecorded by a camera for further analysis.

Scanning electron microscopy

Scanning electron microscopy (SEM) measurements werecarried out with a JSM-6330F microscope (JEOL, Tokyo,

Zhefei Guo et al. Improve chemical stability in HME


Japan). Samples were deposited on the surface of a siliconwafer, then sputter-coated for two cycles with gold.

Physical and chemical stability studies

Stability studies were conducted by placing samples inclosed and open glass vials, which were stored in a climatechamber under a controlled environment, 40C and a rela-tive humidity (RH) of 75%. After 3 months, the crystallinecontent, dissolution, drug content and impurities of thesamples were characterized.

Statistical analysis

The dissolution profiles of the solid dispersions were com-pared using a repeated measures analysis of variance(ANOVA). The influence of polymeric carriers and storageon the dissolution profiles was statistically examined using atwo-way ANOVA (SAS 9.2, SAS Institute Inc., Cary, NC,USA). In all cases, post-hoc comparisons of the means ofindividual groups were performed using Tukeys honestlysignificant difference test. Differences were considered sig-nificant if P < 0.05.

Results

Thermal stability of diflunisal

The thermal stability of DIF was studied by TGA. Theresults were shown in Figure S1. Heating DIF resulted inmelting at 211.8C (onset of melting, while the peak valuewas 214.9C) with the enthalpy of fusion of 123.5 J/g. At theonset of melting, the weight loss was 1.04%, which was con-sistent with the previous study.[20] Once the temperaturewas over 215C, DIF started to decompose seriously and theweight loss increased dramatically.

The results implied that it was impossible to prepareamorphous solid dispersion of DIF at or above the meltingpoint of DIF as such high temperature would definitely leadto severe degradation. Therefore, the process temperatureshould, if possible, be lowered than the melting point ofDIF to get chemically stable solid dispersions.

Melting point depression

The melting point depression of drugpolymer physicalmixtures was studied by DSC. At drug load of 60%, a broad

melting endotherm of DIF presented from 110C to 190C,significantly lower than that of pure DIF (Figure S2). Thiswas possibly due to the strong interaction between DIF andpolymers, which would decrease the chemical potential ofthe drug, and consequently depress the melting point ofdrug.

The interaction parameter () between DIF and polymerwas estimated with the following equation based on FloryHuggins lattice theory.

1 1

11

T T

R

Hln

m

Mmix

Mpure

fusdrug polymer pol

=

+ + yymer2

(1)

where TMmix is the onset melting point of the drug in the

presence of the polymer, TMpure and Hfus represent the

melting point and heat of fusion of pure drug respectively,drug and polymer represent the volume fraction of the drugand polymer, respectively, and m is the ratio of the molecu-lar volume of polymer to that of the drug.[19] According toEq. 1, a plot of 1 1T T H RM

mixMpure

fus drug( ) ( ) ( ) ln1 1 m polymer poly( ) vs mmer2 may yield a linear relationshipat low polymer weight fractions, with the slope equal to .The results in Table 1 showed that all systems had negativevalues of , indicating interactions between DIF andpolymers.

Dissolving process of diflunisal in polymers

HSPM was used to directly illustrate the dissolving processof drug particles in the molten polymer. As shown inFigure 2, the softening of the polymer and the dissolving ofthe drug particles in DIFPVP K30 system took place sim-ultaneously and completed at about 160C. In DIFPVPVA64 and DIFSoluplus systems, once the temperaturereached the Tg of the polymer (106C for PVP VA64 and70C for Soluplus),[21] the polymer was softened to a viscousliquid with drug particles. And the drug particles com-pletely disappeared at 160C. However, in DIFHPMCsystems, solid particles of DIF and HPMC were stillobserved at 160C, which might be due to the relatively highTg of HPMC (170180C).[21]

Table 1 Physical properties used to calculate the FloryHuggins interaction parameters between diflunisal and polymers

Compound MW (g/mol) Density (g/cm3) Molecular volume (cm3/mol) Hfus (kJ/mol)

Diflunisal 250 1.437 174.0 30.90 PVP K30 60 000 1.234 48 622.4 2.3PVP VA64 45 000 1.285 35 026.3 2.0Hydroxypropyl methylcellulose 10 000 1.310 7 578.6 1.8Soluplus 118 000 1.187 96 854.3 1.3



Melt extrusion of diflunisalsolid dispersions

Based on the results of HSPM and DSC, solid dispersionswere prepared at 160C.When the drug load increased from20% to 40%, the maximum torque during HME decreasedin all systems (Table S1). The obvious reduction of torqueindicated that DIF acted as a plasticizer in all systems, andpredominately in DIFPVP K30 systems. The drug contentin the melt extrudates was in the range from 90.1% to102.5% of the initial load (Table S2), and the total unknownimpurities were in the range from 0.21% to 0.30%(Table 2), estimated by HPLC. This indicated that chemi-cally stable solid dispersions of DIF were successfully pre-pared by HME.

Physicochemical properties ofthe solid dispersions

The solid state of DIF, solid dispersions and the correspond-ing physical mixtures was studied by XRD, as shown inFigure 3. The patterns of crystalline DIF showed character-istic diffraction peaks at 2 of 13.8, 14.4, 16.5, 17.0, 23.6,25.8, 26.6, 28.1 and 28.7. However, only peaks correspond-

ing to polymer were observed in patterns of solid disper-sions DIFPVP K30, DIFPVP VA64 and DIFSoluplus.The complete disappearance of DIF crystalline peaks indi-cated that DIF was in an amorphous or molecular state inthe solid dispersions. However, the characteristic peak ofDIF, particularly at 13.8, was distinguished in DIFHPMCsolid dispersions, implying that a small amount of DIF crys-talline remained in the dispersions due to the incompletefusion of DIF in HPMC at 160C.

The morphologies of the solid dispersions and physicalmixtures were observed by SEM, and the results are shownin Figure 4. Solid dispersions, except for DIFHPMCsystem, were homogeneous, which indicated that DIF wascompletely converted into amorphous state and uniformlydistributed in the polymer matrices. While considerableamount of undissolved DIF particles were observed in DIFHPMC solid dispersions, this is consistent with the resultsof HSPM and XRD.

The FTIR spectra of DIF, polymers, solid dispersions andthe corresponding physical mixtures were expanded inregions of 19001600/cm (C=O stretching vibration) and13001100/cm (C-OH stretching vibration) (Figure 5).Since the O-H band of DIF was interfered by aromatic C-H

PVP K30

25 C20% drug

load

160 C20% drug

load

160 C40% drug

load

PVP VA64 Soluplus HPMC

Figure 2 Hot-stage polar microscopy micrographs of phase transition during the heating process of diflunisal and polymer mixtures. Scale bar,200 M.

Table 2 Impurities of solid dispersions (%, n = 3)

DIF load DIFHPMC SD DIFPVP K30 SD DIFSoluplus SD DIFPVP VA64 SD

20% 0.20 0.02 0.30 0.01 0.24 0.01 0.24 0.0130% 0.21 0.02 0.27 0.03 0.21 0.04 0.22 0.0740% 0.23 0.00 0.23 0.04 0.23 0.01 0.24 0.06

DIF, diflunisal; HPMC, hydroxypropyl methylcellulose.



stretch vibrations in the region of 36002800/cm,[22] whichwas also strongly affected by moisture content, this regionwas not selected to explore the intermolecular interactions.For DIFPVP K30 systems (Figure 5a), the spectra of drugshowed a strong band at 1689/cm associated with the C=Ostretch in the carboxylic group, while the polymer showed abroad strong band (1662/cm) due to the C=O stretch in thecyclic amide. These two peaks were clearly observed in thephysical mixture, but only one peak was observed in allDIFPVP K30 solid dispersions. It could not tell whetherthis peak resulted from the broadening or shifting of theband corresponding to carbonyl group in DIF or cyclicamide in PVP K30, but it suggested the formation of inter-molecular hydrogen bonds between DIF and PVP K30 insolid dispersions. In addition, C-OH stretching vibration ofDIF was observed at about 1197/cm, while the polymer didnot show strong absorption in this region. A new peak at1218/cm (20% drug load) arose from the hydrogen-bondedC-OH, showing a shift of about 21/cm relative to the C-OHin pure DIF. This is consistent with the previous reportthat the carboxylic group of indomethacin formed stronghydrogen bonds with the cyclic amide in PVP.[23,24] It wasfound that the shift increased with the increase of polymerconcentration. Since all samples were amorphous as con-firmed by XRD, it implied that the shift did not result fromamorphization but intermolecular interaction. In addition,the peak shift of 20% DIFPVP K30 solid dispersion wasgreater than those of other polymer systems at the samedrug load, suggesting that the hydrogen bonds formedbetween DIF and PVP K30 were the strongest.

In the case of the DIFPVP VA64 systems (Figure 5b),carbonyl peak of pure polymer appeared at 1667/cm, whilecarbonyl peak of DIF was at 1689/cm. For 20% DIFPVPVA64 solid dispersion, a new peak was observed at 1676/cm,which was assigned to the hydrogen-bonded carbonyl. TheC-OH peak of DIF appeared as a shoulder peak at 1203/cmor 1207/cm, respectively, for dispersions with 40% or 30%drug load. Interestingly, as the drug load decreased, thisC-OH peak quickly decreased in intensity and completelydisappeared at 20% drug load. It might be attributed to thehydrogen interaction between the C-OH group of DIF andthe carbonyl group of PVP VA64 (structures are shown inFigure 1).

In the presence of Soluplus (Figure 5c), the changes inthe carbonyl region similarly indicated the formation ofhydrogen bonds at the carbonyl group of DIF or polymer orboth. But the minor shift of C-OH, only 12/cm, indicateda relatively weaker interaction in DIFSoluplus systemscompared with DIFPVP K30 systems.

For DIFHPMC systems (Figure 5d), as HPMC does nothave a carbonyl group, the shifts in both carbonyl regionand C-OH region were attributed to the hydrogen bondsformed at the carbonyl group of DIF. However, the shiftswere only observed in dispersion with 20% drug load, andat higher drug load, the spectra of dispersions were almostidentical to those of physical mixtures. This indicated thatthe intermolecular interaction was very weak in DIFHPMC solid dispersions with 30% and 40% drug load,which was consistent with the results from XRD showingthe presence of drug crystalline particles. Therefore, the

(a)

10

Inte

nsi

tyIn

ten

sity

Inte

nsi

tyIn

ten

sity

20 302-Theta (degree)

40

10 20 302-Theta (degree)

40 10 20 302-Theta (degree)

40

10 20 302-Theta (degree)

40

DIF20% PMPVP K3020% SD30% SD40% SD

DIF20% PMSoluplus20% SD30% SD40% SD

DIF20% PMPVP VA6420% SD30% SD40% SD

DIF20% PMHPMC20% SD30% SD40% SD

(b)

(c) (d)

Figure 3 Powder X-ray diffraction patterns for crystal diflunisal, physical mixtures and solid dispersions using (a) PVP K30, (b) PVP VA64, (c)Soluplus and (d) hydroxypropyl methylcellulose as carriers.



extent of shift as well as the strength of hydrogen bond insolid dispersions followed the sequence of DIFPVPK30 > DIFPVP VA64 > DIFSoluplus > DIFHPMC.

Dissolution behaviours

Due to the poor wettability and solubility, the dissolution ofpure DIF was slow, and only about 20% of DIF was releasedafter 120 min. But its dissolution was dramaticallyimproved by incorporating DIF into hot-melt extrudedsolid dispersion (Figure 6), and this improvement wasdetermined by drug load and polymer type. The dissolutionrate decreased with the increase of drug load (P < 0.05) forall systems, except for DIFHPMC solid dispersion. ForDIFPVP VA64 and DIFPVP K30 dispersions at 20% drugload, drug release was almost completed after 10 min, andno significant difference in dissolution rate was foundbetween the two groups (P > 0.05). However, the dissolu-tion rate of DIF from HPMC and Soluplus matrix was sig-

nificantly slower than from PVP VA64 and PVP K30(P < 0.05), while DIFSoluplus dispersions exhibited theslowest dissolution (P < 0.05). As the drug load increasedto 30%, the dissolution rate of DIFSoluplus dispersionappeared to slow down dramatically (only about 60% DIFwas released after 120 min), significantly slower than that ofthe other three solid dispersions (P < 0.05). Djuric et al.reported that Soluplus could form the polymeric micellesduring the release of itraconazole from the extrudates.[25]

Therefore, the slow release of DIF from DIFSoluplus soliddispersions might be due to the possible formation ofSoluplus micelles, but the actual mechanism needs furtherinvestigation.

For DIFHPMC system, nearly 10% of drug remainedundissolved even after the entire dissolution span of120 min at both 20% and 30% drug loads. It might bebecause part of DIF was still crystalline in DIFHPMC soliddispersion, which was confirmed by XRD, SEM and HSPM.As drug load increased from 20% to 30%, the dissolution

20% PM

PVP

K30

PVP

VA

64So

lup

lus

HPM

C

20% SD 30% SD 40% SD

Figure 4 Scanning electron microscopy photographs of pure polymers, physical mixtures and solid dispersions. Scale bar, 100 m.



profile of DIFHPMC system almost remained unchanged,while the dissolution rate of DIF in other systems appearedto slow down. At 30% drug load, the dissolution rate of DIFfrom HPMC matrix was faster than from PVP K30 matrixeven though DIFPVP K30 system has stronger intermo-lecular hydrogen bond than DIFHPMC system. It is reallyan interesting phenomenon, but the mechanism is stillunclear.

Stability study

Solid dispersions with 20% drug load were subjected toaccelerated stability testing. After they have been stored in

open or closed glass amber bottles at 40C/75% RH for 3months, the dissolution results were listed in Table 3.According to ANOVA results, both polymer carriers andstorage conditions had significantly influence on the disso-lution behaviour of DIF (P < 0.05), and the interactionbetween these two factors was also significant (P < 0.05),which made the mean differences of the accumulativerelease percentages of DIF at 10 min and 30 min amongfour types of solid dispersions not constant before and afterstorage. The following post-hoc comparisons showed thatthe accumulative release percentages of DIF at both 10 minand 30 min barely changed after storage for DIFPVP K30and DIFPVP VA64 solid dispersions (P > 0.05), but signifi-

(a)

(c) (d)

(b)

1750 1700 1650Wave number (cm1)

1689 cm1

1662 cm1

1661 cm1

1672 cm1

1675 cm1

1689 cm1

1197 cm1

1670 cm1

1673 cm1

1672 cm11688 cm1

1686 cm1

1678 cm1

1689 cm11197 cm1

1213 cm1

1202 cm1

1200 cm1

1212 cm1

1216 cm1

1218 cm1

1197 cm1

1197 cm1

1689 cm1

1667 cm1

1676 cm1

1677 cm1

1675 cm1

1200 1150

1750 1700 1650Wave number (cm1)

1200 1150 1750 1700 1650Wave number (cm1)

1200 1150

1750 1700 1650Wave number (cm1)

1200 1150

40% SD30% SD

20% SD

PVP K30

20% PMDIF

40% SD30% SD

20% SD

Soluplus

20% PM

DIF

40% SD30% SD20% SDHPMC

20% PM

DIF

40% SD

30% SD

20% SD

PVP VA6420% PMDIF

Figure 5 Fourier transform infrared spectroscopy of diflunisal, polymers, physical mixtures and solid dispersions showing the stretching regions ofC=O and C-OH, with (a) PVP K30, (b) PVP VA64, (c) Soluplus and (d) hydroxypropyl methylcellulose as carriers.

(a) (b)100

80

60

40

Dis

solu

tio

n (

%)

Dis

solu

tio

n (

%)

20

0

100

80

60

40

20

00 20 40 60 80

Time (min)100 120 0 20 40 60 80

Time (min)100 120

ABCDE

ABCDE

Figure 6 Dissolution profiles of diflunisal and solid dispersions with (a) 20% and (b) 30% drug load, respectively. (A) diflunisal, (B) diflunisalPVPVA64 SD, (C) diflunisalPVP K30 SD, (D) diflunisalSoluplus SD and (E) diflunisalhydroxypropyl methylcellulose SD (n = 3).



cantly decreased after storage for DIFHPMC and DIFSoluplus solid dispersions (P < 0.05).

In addition, the solid dispersions showed no phasetransformation of DIF during stability study, except DIFHPMC solid dispersion which recrystallized, resulting inslower dissolution rate and appearance change. Additionaldiffraction peaks were observed in the diffraction patternof DIFHPMC solid dispersion after storage. This mightbe because the remaining crystalline DIF in extrudates ser-viced as nucleus and facilitated the recrystallization. ForDIFSoluplus solid dispersion, the slowdown in dissolu-tion rate after storage might be attributed to the changeof sample from loose powder to sticky gel, which resultedin the decreasing surface area, retarded disintegrationand poor dispersion of sample in the dissolutionmedium.

Under open-storage conditions, all solid dispersionschanged from loose powders to sticky gels. This might arisefrom the moisture absorption, which had a dramaticdepressing effect on Tg by acting as a plasticizer.[26] There-fore, under high moisture, the samples changed from glassstate to rubbery state and stuck together. Furthermore, theexposure to elevated temperature also helped plasticize theinterface and further increased the molecular mobility.[21]

The differences in the appearance of solid dispersions afterstorage might be due to the differences in their Tgs. The Tgsof DIFPVP K30 (124.3C) and DIFPVP VA64 (80.8C)were higher than those of DIFSoluplus (53.8C) and DIFHPMC (53.5C). When stored at 40C, which was close tothe Tg of DIFHPMC and DIFSoluplus solid dispersions,higher mobility was thus expected to lead to agglomerationand phase separation. For all systems, chemical degradationof DIF was less than 0.5% under accelerated conditions for3 months.

Discussion

Relationship between thermodynamics ofmixing and diflunisalpolymer interactions

According to FloryHuggins lattice theory, the change inthe free energy of mixing could be estimated by the follow-ing equation:[19]

GRT

n ln n ln nmix drug drug polymer polymer drug polymer= + + (2)

where ndrug and npolymer are the mole number of drug andpolymer, respectively, drug and polymer are the volume frac-tion of drug and polymer, respectively, R is the gas constant,T is the absolute temperature, and is the FloryHugginsinteraction parameter representing the interaction betweenpolymer segments and drug molecules. The more negativevalue of indicates stronger interactions.[27] Based on thisequation, the free energy of mixing DIF with polymers aspredicted by melting point depression is shown in Figure 7.All systems showed a negative free energy of mixing nearthe melting temperature, with the most negative value forDIFPVP K30, followed by DIFPVP VA64, DIFHPMCand DIFSoluplus systems sequentially. According tomelting theory, the melting of a pure drug occurs at a tem-perature when the chemical potential of the crystalline drugis equal to that of the molten drug.[9] At a constant pressure,the chemical potential is the partial molar Gibbs freeenergy. Negative free energy of mixing meant the chemicalpotential of drug in the presence of polymer was lower thanthat of the pure amorphous drug. Therefore, the melting ofdrug in the solid dispersions would take place at a lowertemperature than the melting point of the pure drugsubstance.

Table 3 Physicochemical characterization and dissolution evaluation of solid dispersions under accelerated stability conditions

Formulation Conditions, time Appearance Physical state of DIF

% Dissolution

10 min 30 min

20% DIFPVP K30 SD Initial Loose powder Amorphous 95.09 5.57 99.81 0.6540C/75% RH, 3 months, closed Loose powder Amorphous 90.95 3.45 100.08 1.5040C/75% RH, 3 months, open Sticky gel Amorphous

20% DIFPVP VA64 SD Initial Loose powder Amorphous 100.91 0.42 100.68 0.3040C/75% RH, 3 months, closed Loose powder Amorphous 100.75 0.63 100.85 1.3540C/75% RH, 3 months, open Sticky gel Amorphous

20% DIFHPMC SD Initial Loose powder Partially amorphous 81.85 0.58 93.02 0.6640C/75% RH, 3 months, closed Slightly sticky Recrystallization 38.36 7.35 56.50 11.0240C/75% RH, 3 months, open Sticky gel Recrystallization

20% DIFSoluplus SD Initial Loose powder Amorphous 39.87 1.07 72.49 5.0040C/75% RH, 3 months, closed Slightly sticky Amorphous 24.92 1.18 52.29 2.5140C/75% RH, 3 months, open Sticky gel Amorphous

DIF, diflunisal; HPMC, hydroxypropyl methylcellulose; RH, relative humidity.



Based on FloryHuggins lattice theory, the first two termson the right side of Eq. 2 account for the combinatorialentropy contributions, and the third term is an enthalpiccontribution. As the entropy contributions are always nega-tive for mixing, the factors that contribute to the enthalpyof mixing are critical for understanding mixing. The mixingenthalpy is determined by the strength and the number ofthe interactions in the mixture relative to those of the purematerials.[19] In general, intermolecular interactions amongorganic molecules stem from (partial) electron transfer andsharing, electrostatic force, and polarization contributions.Hydrogen bond is a special case of electron transfer andsharing between a hydrogen atom and other specific atomsthat can share a pair of electrons, and it is ready to formbetween drug and polymer towards mixing.[23] In this study,the shifts of C=O and C-OH stretching vibrations in FTIRspectrum were the most obvious for DIFPVP K30 systems,followed by DIFPVP VA64 systems, which indicated thathydrogen bonds formed between DIF and PVP K30 andPVP VA64 were stronger than those for the other twosystems. Consistently, DIFPVP K30 systems had value inmost negative magnitude ( = 2.3), followed by DIFPVPVA64 systems ( = 2.0), indicating stronger interactions(hydrogen bonds) formed in these two systems. Conse-quently, the stronger interaction would result in the lowermelting point of drug in polymers, and the lower processingtemperature during HME.

Manufacturability and diflunisalpolymerinteractions

Thermodynamically, drug substance has a temperature-dependent solubility in polymers. Typically, greater interac-tion between drug and polymer would result in greatersolubility of drug in polymer at a defined temperature, asstudied previously.[28,29] In other words, greater interactionwould lead to lower melting point of drug at a defined drug

load in polymer. The dissolution of drug in polymer wouldonly occur when temperature above the Tg of polymer, astemperature below Tg dissolution would not occur on a rel-evant timescale due to the low mobility of the polymer.[21]

In this study, the obtained solid dispersions were all amor-phous because the drug load was designed below the solu-bility of drug in the polymer at the process temperature toavoid crystalline solid dispersion.

The results from melting point depression showed thatDIFHPMC systems had a more negative value than thatof DIFSoluplus systems, which meant that DIF moleculesshould prefer to be in contact with HPMC segments thanwith Soluplus. However, extrusion of DIFHPMC systemsat 160C resulted in crystalline solid dispersions even whendrug load was only 20%, while DIFSoluplus systemsresulted in amorphous solid dispersions even when drugload was 40%. This is most likely due to the high Tg ofHPMC (170180C),[21] which resulted in non-liquid-likepolymer matrix with low mobility at the processing tem-perature, hindering the contact of DIF with HPMC.

It should be noticed that the HME processing tempera-ture was also lower than the Tg of PVP K30, but the DIFPVP K30 solid dispersions were amorphous even with thedrug load up to 40%. As the torque was considerablyreduced from 4.87 N/m to 1.79 N/m, with the increase ofdrug load from 20% to 30% for DIFPVP K30 systems, itpossibly suggested that DIF itself served as the plasticizer tofacilitate the melt extrusion. During extrusion, PVP K30was in its rubber state, allowing physical and subsequenthydrogen-bonded interactions between drug molecules andpolymer segments.

In comparison, PVP VA64 had a lower Tg (106C) andformed a relatively stronger interaction with DIF, so DIFPVP VA64 systems were expected to have goodmanufacturability. Surprisingly, amorphous solid disper-sions of 30% DIFPVP VA64 were prepared at process tem-perature as low as 140C with screw speed of 250 rpm. Andthese solid dispersions showed similar FTIR spectra anddissolution profiles as that prepared at 160C and 150 rpm(data not shown). This indicated that the solubility of DIFin PVP VA64 was over 30% at 140C, and the higher screwspeed provided greater shear to thin the boundary layer andenhance the dissolution.

Conclusions

Drugpolymer interaction was utilized to depress HMEprocessing temperature for preventing the thermal degrada-tion of heat-sensitive drug. The melting point of DIF wasdramatically depressed by the miscible polymers. Based onthese results, amorphous solid dispersions of DIF were suc-cessfully hot-melt extruded at 160C, 55C lower than thenormal melting point of DIF. The total impurities were less

0.0000PVP K30PVP VA64HPMCSoluplus0.0001

0.0002

0.0003

0.0004

0.0 0.2 0.4Polymer

GM

/RT

0.6 0.8 1.0

Figure 7 Free energy of mixing diflunisal with polymers using meltingpoint depression measurements.



than 0.3%, indicating that the thermal degradation of DIFduring HME was effectively minimized. The DIF in soliddispersions was still amorphous after storage at 45C and75% RH for 3 months except for DIFHPMC system.Hydrogen bonds between DIF and polymers were suggestedto be the major contributor to the improved chemical andphysical stability. The dissolution of DIF in the extrudedsolid dispersions was remarkably improved, especially at20% drug load. This technique based on drugpolymerinteraction to hot-melt extrude chemically stable amor-phous solid dispersions provides an attractive opportunityfor the development of heat-sensitive drugs by HME.

Declarations

Funding

This work was funded by the Fundamental Research Fundsfor the Central Universities (No. 12ykpy08), the NationalNatural Science Function of China (No. 51103184),National Science & Technology Pillar Program (No.2012BAI35B02), and International Cooperation andExchanges Program of China (No. 2008DFA31080).

References

1. Hancock BC, Zografi G. Characteris-tics and significance of the amorphousstate in pharmaceutical systems. JPharm Sci 1997; 86: 112.

2. Puri V et al. Barrier coated druglayered particles for enhanced perfor-mance of amorphous solid dispersiondosage form. J Pharm Sci 2012; 101:342353.

3. Janssens S, Van den Mooter G.Review: physical chemistry of soliddispersions. J Pharm Pharmacol 2009;61: 15711586.

4. Qian F et al. Drug-polymer solubilityand miscibility: stability considerationand practical challenges in amorphoussolid dispersion development. J PharmSci 2010; 99: 29412947.

5. Yu L. Amorphous pharmaceuticalsolids: preparation, characterizationand stabilization. Adv Drug Deliv Rev2001; 48: 2742.

6. Albers J et al. Mechanism of drugrelease from polymethacrylate-basedextrudates and milled strands pre-pared by hot-melt extrusion. Eur JPharm Biopharm 2009; 71: 387394.

7. Suwardie H et al. Rheological study ofthe mixture of acetaminophen andpolyethylene oxide for hot-melt extru-sion application. Eur J PharmBiopharm 2011; 78: 506512.

8. Frank KJ et al. Amorphous soliddispersion enhances permeation ofpoorly soluble ABT-102: true super-saturation vs. apparent solubilityenhancement. Int J Pharm 2012; 437:288293.

9. Ghosh I et al. Key considerations foroptimization of formulation andmelt-extrusion process parameters fordeveloping thermosensitive com-pound. Pharm Dev Technol 2012; 17:502510.

10. DiNunzio JC et al. Fusion productionof solid dispersions containing a heat-sensitive active ingredient by hot meltextrusion and Kinetisol dispersing.Eur J Pharm Biopharm 2010; 74: 340351.

11. Liu X et al. Improving the chemicalstability of amorphous solid disper-sion with cocrystal technique by hotmelt extrusion. Pharm Res 2012; 29:806817.

12. Childs SL et al. Screening strategiesbased on solubility and solutioncomposition generate pharmaceuti-cally acceptable cocrystals ofcarbamazepine. CrystEngComm 2008;10: 856864.

13. Lakshman JP et al. Application of meltextrusion in the development of aphysically and chemically stable high-energy amorphous solid dispersion ofa poorly water-soluble drug. MolPharm 2008; 5: 9941002.

14. Ghosh I et al. Comparison of HPMCbased polymers performance as carri-ers for manufacture of solid disper-sions using the melt extruder. Int JPharm 2011; 419: 1219.

15. Verreck G et al. Hot stage extrusion ofp-amino salicylic acid with EC usingCO2 as a temporary plasticizer. Int JPharm 2006; 327: 4550.

16. Abu-Diak OA et al. Understanding theperformance of melt-extruded poly-

(ethylene oxide)-bicalutamide soliddispersions: characterisation ofmicrostructural properties usingthermal, spectroscopic and drugrelease methods. J Pharm Sci 2012;101: 200213.

17. Qi S et al. Characterisation of soliddispersions of paracetamol andEUDRAGIT E prepared by hot-meltextrusion using thermal, micro-thermal and spectroscopic analysis. IntJ Pharm 2008; 354: 158167.

18. Vogt FG et al. 17O solid-state NMR asa sensitive probe of hydrogen bondingin crystalline and amorphous solidforms of diflunisal. Mol Pharm 2013;10: 34333446.

19. Marsac P et al. Theoretical and practi-cal approaches for prediction of drug-polymer miscibility and solubility.Pharm Res 2006; 23: 24172426.

20. Wassel AA. Thermal stability of someanti-inflammatory pharmaceuticaldrugs and determination of purity. JPhys Chem A 2010; 1: 18.

21. DiNunzio JC et al. Melt extrusion. In:Williams RO III, ed. FormulatingPoorly Water Soluble Drugs. New York,NY: Springer, 2012: 311362.

22. Martnez-Ohrriz MC et al. Solid dis-persions of diflunisal-PVP: polymor-phic and amorphous states of thedrug. Drug Dev Ind Pharm 2002; 28:717725.

23. Marsac P et al. Estimation of drug-polymer miscibility and solubility inamorphous solid dispersions usingexperimentally determined interactionparameters. Pharm Res 2009; 26: 139151.



24. Taylor L, Zografi G. Spectroscopiccharacterization of interactionsbetween PVP and indomethacin inamorphous molecular dispersions.Pharm Res 1997; 14: 16911698.

25. Djuric D et al. Characterization ofpolymeric micelles from solid solu-tions with a polyvinyl caprolactam-polyvinyl acetate-polyethylene glycolgraft copolymer and itraconazole.38th Annual Meeting & Expositionof the Gontrolled Release Society,Maryland, 2011. Available at: http://www.pharma-ingredients.basf.com/Documents/ENP/Poster/EN/GENPMD343.pdf.

26. Rumondor ACF et al. Phase behaviorof poly(vinylpyrrolidone) containing

amorphous solid dispersions in thepresence of moisture. Mol Pharm2009; 6: 14921505.

27. Lin DX, Huang YB. A thermal analysismethod to predict the complete phasediagram of drug-polymer solid disper-sions. Int J Pharm 2010; 399: 109115.

28. Sun Y et al. Solubilities of crystallinedrugs in polymers: an improvedanalytical method and comparisonof solubilities of indomethacin andnifedipine in PVP, PVP/VA, and PVAc.J Pharm Sci 2010; 99: 40234031.

29. Tao J et al. solubility of small-molecule crystals in polymers:d-mannitol in PVP, indomethacin inPVP/VA, and nifedipine in PVP/VA.Pharm Res 2009; 26: 855864.

Supporting InformationAdditional Supporting Information maybe found in the online version of thisarticle at the publishers web-site:

Figure S1 Thermal gravimetric analysisand differential scanning calorimetry pro-files of crystalline diflunisal.Figure S2 Differential scanning calorim-etry profiles of crystalline diflunisaland physical mixtures of diflunisal andpolymers.Table S1 Maximum torque (N/m)during hot-melt extrusion.Table S2 Drug content of solid disper-sions (%, n = 3).



Drug Polym Interactions

Documents

Transcript of Drug Polym Interactions