Effect of Temperature and Moisture on the Physical Stability of Binary and Ternary Amorphous Solid Dispersions of Celecoxib

Journal of Pharmaceutical Sciences xxx (2016) 1-11

Contents lists available at ScienceDirect

Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org

Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Effect of Temperature and Moisture on the Physical Stability of Binary and Ternary Amorphous Solid Dispersions of Celecoxib Tian Xie, Lynne S. Taylor* Department of Industrial and Physical Pharmacy, College of Pharmacy, Purdue University, West Lafayette, Indiana 47907

a r t i c l e i n f o

a b s t r a c t

Article history: Received 27 February 2016 Revised 26 May 2016 Accepted 9 June 2016

The effectiveness of different polymers, alone or in combination, in inhibiting the crystallization of celecoxib (CEX) from amorphous solid dispersions (ASDs) exposed to different temperatures and relative humidities was evaluated. It was found that polyvinylpyrrolidone (PVP) and PVP-vinyl acetate formed stronger or more extensive hydrogen bonding with CEX than cellulose-based polymers. This, combined with their better effectiveness in raising the glass transition temperature (Tg) of the dispersions, provided better physical stabilization of amorphous CEX against crystallization in the absence of moisture when compared with dispersions formed with cellulose derivatives. In ternary dispersions containing 2 polymers, the physical stability was minimally impaired by the presence of a cellulose-based polymer when the major polymer present was PVP. On exposure to moisture, stability of the CEX ASDs was strongly affected by both the dispersion hygroscopicity and the strength of the intermolecular interactions. Binary and ternary ASDs containing PVP appeared to undergo partial amorphouseamorphous phase separation when exposed 94% relative humidity, followed by crystallization, whereas other binary ASDs crystallized directly without amorphouseamorphous phase separation. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

Keywords: amorphous crystallization polymers stabilization solubility

Introduction Drugs formulated as the amorphous form can provide advantages over their crystalline counterparts by offering increased kinetic solubility and a faster dissolution rate, contributing to improved absorption in vivo.1,2 However, such improvements in delivery are inevitably accompanied with an increased risk of product instability, on account of the intrinsic tendency of an amorphous compound to crystallize during storage. A polymer is typically mixed with the amorphous drug at the molecular level to inhibit solid-state crystallization. The origin of the stabilizing effect of polymers is multifaceted. Polymers are thought to reduce the molecular mobility of the amorphous drug by increasing the glass transition temperature (Tg) of the system,3 typically referred to as an antiplasticization effect. However, other studies have revealed that crystallization inhibition by a polymer cannot always be rationalized based on Tg changes.4-6 Polymers may also form specific interspecies interactions with the drug, such as hydrogen bonding5,7,8 or ionic interactions.9,10 The strength of such

This article contains supplementary material available from the authors by request or via the Internet at http://dx.doi.org/10.1016/j.xphs.2016.06.017. * Correspondence to: Lynne S. Taylor (Telephone: 765-714-2808; Fax: 765-4946545). E-mail address: [email protected] (L.S. Taylor).

interactions appears to have some correlation with the relative effectiveness of the polymers as crystallization inhibitors.11 Crystalline solids adsorb water by a surface adsorption mechanism.12 Amorphous solids are more hygroscopic than their crystalline counterparts due to their disordered structure and higher free volume. In addition to surface adsorption, water can penetrate into the bulk of amorphous materials. The moisture absorbed can be detrimental to the stability of amorphous solid dispersions (ASDs). Water (Tg ¼ 137 C)13 can effectively reduce the Tg of the ASD and thereby increase the molecular mobility.14,15 Water can also irreversibly disrupt the drugepolymer interactions by competitively forming hydrogen bonds with a hydrophilic polymer.16 Both factors may induce amorphouseamorphous phase separation (AAPS)17,18 in the ASDs with subsequent crystallization that preferentially occurs in the drug-rich domains.19 Alternatively, the increased molecular mobility resulting from absorbed water can lead to direct crystallization of an amorphous drug.15,20 While it is obviously vital to ensure that the drug remains amorphous during a typical shelf life of 3-5 years,21 it is of equal importance to inhibit crystallization during dissolution over pharmaceutically relevant timeframes. Unfortunately, choosing the optimum inhibitor for either crystallization pathways is largely empirical. Furthermore, the polymer that is the best solid-state inhibitor may be ineffective as a solution crystallization inhibitor and vice versa.22-25 For example, it was observed that polyacrylic

http://dx.doi.org/10.1016/j.xphs.2016.06.017 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

2

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

Figure 1. Chemical structures of CEX (a), PVP (b), PVP/VA (c), HPMCAS (d), and HPMC (e).

acid was very effective at inhibiting the crystal growth of acetaminophen from supercooled liquids7,26 and hence prevented crystallization from ASDs, but was ineffective at preventing either nucleation or crystal growth from aqueous solutions.27 Therefore, an important advance in ASD formulation could be to incorporate combinations of polymers, chosen to maximize crystallization inhibition in both the solid and solution phases.28-30 Before this strategy can be implemented, however, it is necessary to determine the miscibility and stability of the ternary systems. If the polymers are miscible, it should be relatively straightforward to determine if a 1-phase amorphous system is produced when the drug is added. Furthermore, if drugepolymer interactions are indeed important in influencing the stability of the system to crystallization, it becomes essential to elucidate the various interspecies interactions within the blend and the resultant physical stability in ternary systems. To the best of our knowledge, these issues have not been extensively probed to date. Herein, we evaluated the physical stability of celecoxib (CEX), a poorly soluble anti-inflammatory agent,31 when formulated in ASDs with different polymers and exposed to high-stress storage conditions of elevated temperature and relative humidity (RH). Previously, we reported that the dissolution performance of high drug-loading ASDs of CEX formulated with polyvinylpyrrolidone (PVP) was improved when 20% of the PVP was replaced by an effective solution crystallization-inhibiting polymer either hydroxypropyl methylcellulose (HPMC) or HPMC acetate succinate (HPMCAS).32 It was therefore also of interest to evaluate the miscibility and physical stability of ternary ASDs relative to the corresponding binary ASDs.

Materials CEX was purchased from Attix Pharmaceuticals (Toronto, Ontario, Canada). PVP (K29/32: Mw 58,000 g mol1) was purchased from ISP Technologies, Inc (Wayne, NJ). PVP (K12: MW 2000-3000 g mol1), and PVP-vinyl acetate (PVP/VA) (Kollidon VA64, Mw 45,00047,000 g mol1) was provided by the BASF Corporation (Ludwigshafen, Germany). HPMCAS (Type AS-MF: Mw 17,000 g mol1) and HPMC (Type 606: Mw 35,600 g mol1) were supplied by Shin-Etsu Chemical Company (Tokyo, Japan). The molecular structures of the drug and polymers are shown in Figure 1.

Methods Preparation of Bulk Amorphous Materials CEX and the polymer(s) at different dry weight ratios were dissolved in a mixture of ethanol and dichloromethane. Solvent removal was achieved by rotary evaporation. The ASDs were subsequently dried in a vacuum oven overnight to remove any residual solvent. Samples were then ground using a mortar and pestle, and sieved to obtain a particle size fraction of 106-250 mm and stored in a desiccator containing phosphorus pentoxide at room temperature (RT) until use. IR Spectroscopy Solutions of drug and polymer mixtures at different dry weight ratios or single component polymers were dissolved in a mixture of ethanol and dichloromethane. The solutions were then dipped onto thallium bromoiodide (KRS-5) optical crystals and rotated on a KW4A spin coater (Chemat Technology, Inc., Northridge, CA). IR spectra of the resulting thin films were obtained on a Bruker Vertex 70 (Bruker, Billerica, MA). Sixty-four scans were collected in transmission mode with a 4 cm1 resolution for each sample over the wave number range from 4000 to 500 cm1. Dry air was purged into the sampling and optical compartment to prevent spectral interference from water vapor. Pure crystalline CEX was used as is. Amorphous CEX was obtained by melting quenching from pure crystalline CEX as described previously.32 The pure crystalline or amorphous CEX was placed on an attenuated total reflectance attachment (High Temperature Golden Gate ATR; Specac, Inc, United Kingdom), and the IR spectra were obtained using the same Bruker instrument. The Opus 7.2 software (Bruker) was used to analyze the spectra. Spectra were converted into absorbance, normalized, and offset for purposes of plotting and comparison between samples. Storage Conditions and Powder X-Ray Diffraction (PXRD) ASDs of CEX were stored in 20-mL open scintillation vials in a desiccator and then subjected to different environmental conditions for long-term stability studies: (1) 80 C/0% RH

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

(Drierite®), (2) 60 C/0% RH (Drierite®), (3) 40 C/0% RH (Drierite®), (4) RT (~20 C)/56% RH (sodium bromide); (5) RT/75% RH (sodium chloride), (6) RT/94% RH (potassium nitrate), and (7) 40 C/75% RH (sodium chloride). PXRD patterns were taken periodically using a Rigaku SmartLab diffractometer (Rigaku Cooperation, The Woodlands, TX). The patterns were collected in step scan mode from 5 to 35 at 4 /min. The tube voltage and current were 44 kV and 40 mA, respectively. A silicon standard was used to calibrate the instrument. Thermal Analysis Thermal analysis was carried out using a TA Q2000 differential scanning calorimetry (DSC) with a cooling refrigerator system (TA Instruments, New Castle, DE). Indium and tin were used for calibration of temperature, and indium was used for calibration of enthalpy. Dry nitrogen was purged at 50 mL min1. A total of 3-6 mg of the samples were weighed into an aluminum Tzero pan and sealed using a hermetic Tzero lid to which a pinhole was added. The Tg was determined by heating the sample at 20 C min1 to approximately 30 C above Tg, followed by cooling and reheating at 20 C min1 to 180 C, repeating the cycle twice. The onset Tg of the second heating scan was reported. To investigate the impact of moisture, 3-6 mg binary ASDs (6:4) were placed in open Tzero pans and stored at 75% RH or 94% RH at RT. After 3 days of exposure, the pans were hermetically sealed. The samples were then equilibrated to 20 C, heated at 20 C min1 to 100 C, followed by equilibration to 20 C and reheating at 20 C min1 to 100 C twice. The onset Tg from both the first and second heating scan was reported. Dynamic Vapor Sorption Moisture sorption isotherms of ASDs and pure components were measured by a symmetrical gravimetric analyzer (SGA-100; VTI Corporation, Hialeah, FL). Samples were dried at 45 C with dry nitrogen purged into the sorption analyzer before exposure to increasing RH. The equilibrium criterion for the drying step was <0.01% w/w change within 2 min with a maximum drying time of 60 min. The sample was then exposed to RH from 5% to 95% with a 10% step increase at 25 C. The maximum time for equilibration in each step was 180 min.

3

form of CEX thus appears weaker than that in the crystalline form,33 since a lower peak frequency of the N-H group is typically associated with a shorter hydrogen bond distance.34 In ASDs of CEX with PVP, a new peak emerged at 3196 cm1 and increased in intensity at the expense of the 3268 cm1 peak as the polymer loading increased (Fig. 2a). Correspondingly, the C¼O peak of PVP, which appears at 1682 cm1 in the pure dry polymer, developed a shoulder at 1661 cm1 (the drug has no absorbance at this wave number), which increased in intensity as the drug-topolymer ratio increased. This clearly indicates the formation of hydrogen bonding between the NH2 group of CEX and the C¼O of PVP. PVP/VA has 2 types of C¼O group, pyrrolidone and acetate, with characteristic peaks at 1684 and 1737 cm1, respectively. Similar to the case of CEX:PVP, the polymer C¼O peak at 1684 cm1 developed a shoulder at 1662 cm1 when PVP/VA was mixed with CEX, while the acetate carbonyl peak showed minimal change at all but the highest drug loadings (Fig. S2). Correspondingly, the NH peak of CEX shifted from 3268 cm1 to approximately 3198 cm1. These spectral changes indicate that the hydrogen bonding strength between the C¼O group of the vinylpyrrolidone monomer of PVP/VA and the NH2 group of CEX was comparable to that observed in CEX:PVP dispersions. However, for a given drug loading, it appears that a larger fraction of drug molecules was involved in hydrogen bonding in the case of CEX:PVP-K29/32 than for CEX:PVP/VA. This can be presumably attributed to the dilution effect caused by the presence of the vinylacetate monomer in PVP/VA.35 In CEX:HPMCAS, the NH peak of CEX at 3268 cm1 gradually shifted toward 3237 cm1 as the polymer loading increased, whereas the HPMCAS ester carbonyl peak17 shifted from 1743 to 1734 cm1 (Fig. 2b) as the drug loading increased. The smaller magnitude of the shifts for the relevant hydrogen bonding groups relative to the dispersions with PVP or PVP/VA indicate a much weaker interaction between CEX and HPMCAS. HPMC does not have any carbonyl groups to act as a hydrogen bond acceptor to form hydrogen bond with the NH2 group of CEX but does have other acceptor groups, namely the oxygen atoms of the hydroxyl or ether groups. However, it is difficult to see clear spectral evidence of intermolecular interactions (Fig. S3). Hence, if drugepolymer interactions are present, the strength appears to be similar to those found in amorphous CEX based on the relatively minor changes in N-H stretching region observed between pure amorphous drug and dispersion.

Crystal Growth Rate Measurements Crystalline CEX and the polymer(s) were mixed in a cryogenic mill (6750 freezer mill; Spex SamplePrep, Metuchen, NJ). The physical mixture was then melted between 2 cover slips on a hot plate and quench cooled to RT on a flat metal surface. The cover slips were stored at 80 C until small nuclei were formed and became visible. The increase in diameter of the crystal nuclei with time on a hot stage set at the temperature of interest was measured using a polarizing microscope as described previously.23 Results Assessment of DrugePolymer Intermolecular Interactions Using IR Spectroscopy Binary CEX-Polymer Dispersions Crystalline CEX shows a doublet at 3322 and 3225 cm1, arising from N-H stretching vibrations associated with the asymmetric and symmetric motions, respectively, of the NH2 of the sulfonamide group (Fig. S1). These peaks shifted to 3351 and 3268 cm1 for amorphous CEX. The hydrogen bonding in the pure amorphous

Binary Polymer Blends IR spectroscopy has also been widely used to evaluate miscibility and interactions in polymer blends.36 For the PVP:HPMCAS films, clear signs of miscibility between the 2 polymers could be observed based on evidence of hydrogen-bonding interactions, in agreement with previous reports.37 The PVP carbonyl peak at 1682 cm1 developed a shoulder at 1643 cm1, assigned to the C¼O of PVP hydrogen bonded to the carboxylic acid OH group of HPMCAS (Fig. 3). This shoulder occurred at a lower wave number than that observed in the CEX:PVP films, indicating that the hydrogen bonding between the 2 polymers was stronger than that between the drug and PVP. However, the fact that the peak at 1682 cm1 persists and was dominant in the binary polymer blend spectra indicates that not all the PVP monomers were involved in hydrogen bonding with HPMCAS even when HPMCAS was present at high weight fractions. This is consistent with a low overall fraction of carboxylic acid groups in HPMCAS.35 Ternary Mixtures From the results presented previously, it is clear that CEX forms stronger interactions with PVP relative to with HPMCAS but that

4

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

Figure 2. FTIR spectra of (a) CEX:PVP and (b) CEX:HPMCAS spin-coated films (from bottom to top: pure amorphous CEX [orange], 9:1 [green], 7:3 [brown], 5:5 [cyan], 3:7 [pink], 1:9 [red], and pure polymer [orange], respectively) showing the wave number range from 3100 to 3500 cm1 and from 1550 to 1850 cm1.

PVP can form a favorable interaction with HPMCAS. For ternary mixtures, where HPMCAS was present as a minor component (10 wt. % or less), the spectra (Fig. S4) were similar to those obtained from a binary mixture of CEX and PVP, and no shoulder at 1643 cm1 (characteristic of PVP-HPMCAS interactions) can be observed. It is unclear whether hydrogen bonding between PVP and HPMCAS exists in the ternary systems and whether it will form at the cost of that between CEX and PVP. To further evaluate this, the HPMCAS loading was increased up to 40%, and it was observed that the hydrogen bonding between CEX and PVP persisted in this dispersion (Fig. S5). Evidence of hydrogen bonding between PVP and HPMCAS eventually surfaced when the HPMCAS loading was further increased to 60% (Fig. S6), and the extent of the interaction appeared less pronounced in the presence of CEX relative to that in the absence of the drug. This suggests that hydrogen bonding between CEX and PVP is favorable relative to other interaction patterns and thus persists in the ternary mixture and is not substantially disrupted by the presence of HPMCAS. Impact of Absorbed Water on DrugePolymer Interactions On exposure to moisture (94% RH/RT), the peak at 3196 cm1 of CEX:PVP-K29/32 films decreased in intensity, whereas the peak at 3268 cm1 increased in intensity. Correspondingly, there was a slight but discernable decrease in C¼O shoulder at 1661 cm1.

Example spectra for a 3:7 CEX:PVP dispersion are shown in Figure 4, and trends were similar for other drug:polymer ratios in terms of the spectral regions showing changes and the types of changes being observed. After 18 days of storage, crystallization was observed in the 7:3 CEX:PVP dispersion, both from changes in the spectra and from the loss of transparency of the spin-coated film by visual inspection. These results suggest that water irreversibly disrupts some of the hydrogen-bonding interactions between PVP and CEX and possibly induced some extent of AAPS before crystallization.38 In contrast, CEX:HPMCAS and CEX:PVP/VA films appeared resistant to AAPS and crystallization as they showed no notable changes in the spectra following exposure to moisture over 18 days (Figs. S7 and S8). After storage for more than a month, a minor level of crystallization was observed on the edge of the substrates in some high drug loading samples. The changes in spectra of the ternary films were similar to those observed for the binary CEX:PVP films owing to the fact that PVP was present in a larger fraction than HPMCAS (Fig. S9). Isothermal Moisture Sorption As shown in Figure 5, pure amorphous CEX was very hydrophobic with only 2 wt. % water absorbed at 95% RH. The pure

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

5

Figure 3. FTIR spectra of PVP:HPMCAS spin-coated films showing the wave number range from 1600 to 1800 cm1.

polymers differ substantially in hygroscopicity, with the order from highest to lowest moisture sorption being PVP > PVP/VA > HPMC > HPMCAS. Moreover, the amount of water absorbed by PVP was minimally affected by the molecular weight grade as observed previously.39 The ASDs show the same rank order of hygroscopicity as the corresponding pure polymer with the exception of the CEX:PVP/VA ASD in which moisture sorption was largely suppressed, and similar in profile to that of CEX:HPMC ASD (Fig. 5). Furthermore, it can be noted that the binary ASDs, in particular PVP-containing systems, showed a notable increase in moisture sorption when the RH was increased to above 75%, with the exception of the CEX:HPMCAS ASD. Replacing a small portion of PVP-K29/32 with a less

hygroscopic cellulosic polymer did result in slight reductions in water uptake. However, the overall pattern of the profile of the ternary ASDs was similar to that of the corresponding binary CEX:PVP-K29/32 ASDs. Similar profiles were observed in PVP-K12 containing ternary ASDs (data not shown). Thermal Analysis Despite some practical limitations,40 DSC analysis remains the most frequently used method to probe the mixing state of a drug with a polymer.29,41,42 For binary ASDs of the drug with different polymers, a single Tg value was obtained for each composition at a value intermediate to those of the pure components (Fig. 6),

Figure 4. FTIR spectra of pure PVP (blue), CEX:PVP 3:7 unexposed (red), 1 day (green), 18 days (purple), and amorphous CEX (orange) spin-coated films showing the wave number range from 3100 to 3500 cm1 and from 1550 to 1800 cm1. After 18 days of storage, crystallization was observed.

6

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

Figure 5. Moisture sorption isotherms of pure amorphous CEX and polymers (left) and ASDs of CEX (right).

suggesting that the drug is miscible with these polymers. PVP-K12 and PVP/VA have lower Tg values than HPMCAS and HPMC, but apparently more effectively raised the Tg values of the CEX ASDs. Polymerepolymer blends are frequently largely immiscible because of the low entropy of mixing for 2 large molecules.36 Consequently, miscibility between polymers can only be anticipated when they can form specific interspecies interactions and hence have a favorable enthalpic contribution to the free energy of mixing.43 For binary polymer blends of PVP and HPMCAS, a single Tg value was obtained for each composition (Fig. S10). This agrees with the IR results that PVP can form strong hydrogen bonds with HPMCAS (Fig. 3) and that the 2 polymers are miscible. With 3 components, the phase behavior in the solid state immediately becomes very complex. Understanding the miscibility of the ternary system is of crucial importance, since it is desirable

that the solid-stateestabilizing polymer is in the same amorphous phase as the drug. The IR results suggest that CEX has a higher affinity for PVP than HPMCAS in terms of hydrogen-bonding tendency, which is not impaired by competition from HPMCAS for hydrogen bonding with PVP. Hence, a miscible ternary system is expected. This supposition is further supported by the single Tg values obtained for CEX:PVP:HPMCAS ASDs (Table S1). Moreover, the ternary ASD's Tg values were close to but slightly lower than those of the CEX:PVP binary ASDs as expected. To evaluate the impact of moisture on the phase behavior and molecular mobility, CEX ASDs were exposed to 94% RH at RT in open pans for 3 days and then hermetically sealed. In the initial heating scan, binary ASDs of CEX with PVP/VA or HPMCAS or HPMC showed a single Tg (Fig. 7), implying that water uptake did not result in AAPS. In contrast, PVP ASDs showed 2 distinct Tg values.

Figure 6. Tg values of ASDs of CEX with a single polymer.

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

7

Figure 7. Thermograms of binary ASDs (6:4) showing the first (left) and the second (right) heating scan. Samples were hermetically sealed after 3 days of exposure to (a) 94% RH and (b) 75% RH before the experiments. Green: CEX: PVP-K12, purple: CEX:PVP-K29/32, blue: CEX:PVP/VA, black: CEX:HPMC, and red: CEX:HPMCAS. The arrows indicate evidence of 2 glass transitions in some samples.

This is consistent with the IR results shown previously, suggesting that the presence of absorbed water resulted in partial demixing of the drug with PVP. While the exact composition in these amorphous phases cannot be quantified, it is nevertheless surmised that the Tg event at lower temperature is associated with the glass transition of a drug-rich domain, and vice versa. During reheating, however, only a single Tg was observed in all binary dispersions including CEX:PVP. A possible explanation is that the initially separated phases in CEX:PVP remixed during heating.16 At 75% RH, the water uptake by PVP ASDs was substantially reduced relative to that at 94% RH. Correspondingly, the moisture-induced AAPS appeared less pronounced in the PVP-K12 ASD. No evidence of AAPS was observed in the binary PVP-K29/32 ASD or any other binary ASDs. Crystal Growth Rate To quantitatively evaluate the stabilization effect of different polymer(s), crystal growth rate measurements in the presence and absence of the polymers were determined at elevated temperatures. As shown in Figure 8, CEX shows the highest growth rate in the absence of any polymer. Crystal growth was much faster in the presence of HPMCAS than that in the presence of PVP at an equivalent weight percentage of polymer. Increasing the molecular weight of PVP minimally reduced the growth rate relative to the

lower molecular weight grade. The crystal growth rate of CEX in the presence of PVP/VA was slightly faster than that in the presence of PVP. This may be because more vinylpyrrolidone carbonyl groups are present in PVP than those in PVP/VA at the same polymer loading,35 enabling more drugepolymer interactions to be formed for a given weight fraction of polymer. Replacing a minor proportion of PVP with HPMCAS led to a small increase in crystal growth rate relative to the corresponding binary PVP system, although the growth rate remained substantially lower than that of the CEX:HPMCAS dispersion. Physical Stability During Different Storage ConditionsdTemperature Versus Humidity Effects Figure 9a shows PXRD patterns of the binary and ternary ASDs stored at 80 C/0% RH. At 80% drug loading, the most stable ASDs were the binary CEX:PVP-K29/32 and CEX:PVP/VA ASDs as well as the ternary ASD with PVP and a small amount of HPMCAS or HPMC, which showed signs of crystallization only after ~120 days of storage. In contrast, CEX:HPMCAS and CEX:HPMC ASDs crystallized the fastest, showing evidence of crystallinity after 6 days. In addition, the results indicate that the physical stability of ternary ASDs could be largely predicted by considering the corresponding binary ASD with a similar composition. The crystal

8

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

Figure 8. Crystal growth rate of CEX in the absence and presence of polymers.

growth rate measurements appear to be predictive the relative crystallization kinetics of bulk CEX ASDs, which might potentially substantially reduce the time and cost associated with long-term physical stability studies of ASD formulations, since these measurements can be performed relatively quickly. For ASDs with 60% drug loading stored at RT/94% RH (Fig. 9b), CEX:PVP/VA was the most stable binary ASD, remaining amorphous for up to 150 days and crystallizing thereafter. CEX:HPMCAS was stable for up to 115 days before crystallizing. The binary ASDs with HPMC or PVP and the ternary ASDs where PVP was the major polymer were much less stable, with some diffraction peaks being observed within a month. At 40 C/75% RH where both the temperature and humidity were elevated relative to ambient conditions, crystallization occurred at an earlier time point in all systems. The CEX:PVP/VA 8:2 ASD was the most stable and remained amorphous for up to 4 months. The CEX:HPMCAS dispersion was stable for up to 33 days, and crystallization was detected thereafter. Interestingly, at the same drug loading, CEX:PVP-K29/32 and the corresponding ternary ASD outperform CEX:HPMCAS, remaining amorphous for up to 70 days. Discussion Previously, we observed that both HPMC and HPMCAS are excellent inhibitors of CEX crystallization from supersaturated solutions and have demonstrated that using combinations of PVP and HPMCAS in dispersion can lead to improved release at high drug loading.32 Investigation of the physical stability of ternary CEX ASDs relative to the corresponding binary ASDs is the next logical step in evaluating the utility of using polymer combinations. In the present study, it was found that replacing a small fraction of PVP with HPMCAS or HPMC does not substantially impact the crystallization kinetics, and the ternary dispersion showed a similar stability to the binary CEX:PVP dispersion. This is an important observation, since it lends support to our contention that ASDs containing binary polymer combinations may be useful to optimize performance in different scenarios (e.g., storage and dissolution).

In terms of differentiation between polymers, we found that PVP and PVP/VA were more effective in raising the Tg of CEXcontaining ASDs than the cellulosic derivatives, which may contribute to the observed difference in the crystallization kinetics under nonhumid storage conditions (Fig. 9a), whereby they were more effective than the cellulose derivatives. This is perhaps, at least intuitively, consistent with the existence of much stronger hydrogen-bonding interactions between CEX and PVP or PVP/VA than those between CEX and the cellulose derivatives, demonstrated in the IR studies, which may restrict the diffusive motion and hence self-association of the drug molecules in the dispersions. Similarly, Miyazaki et al.7 found that polyacrylic acid provided better physical stabilization for amorphous acetaminophen than PVP by forming a stronger hydrogen-bonding interaction with the drug and leading to a larger decrease in molecular mobility, as indicated by the longer enthalpy relaxation times, despite a similar Tg value. Several other examples have been presented in the literature linking drugepolymer intermolecular interactions to physical stability.44 Importantly, the addition of a small amount of HPMCAS was not able to disrupt the drugePVP intermolecular interactions and hence did not substantially impact the crystallization kinetics. While the ability of the different polymers to inhibit crystallization from ASDs after storage at high temperature can be largely explained by consideration of intermolecular interactions and Tg values, the crystallization trends after storage under humid conditions appear to be more complex (Fig. 9b). Table 1 lists the Tg values of ASDs (6:4) before and after exposure to different RH conditions for 3 days at RT. It is clear that the Tg values (i.e., a measure of molecular mobility) alone cannot explain the disparity in ASD stability at 94% RH, since these values are similar for all the binary ASDs (except PVP-K12 ASD), while the crystallization kinetics are quite different, with the PVP-containing dispersions being the most susceptible to crystallization (Fig. 9b). However, if the reduction in Tg is plotted (Fig. 10), it is apparent that PVP dispersions show a larger reduction in Tg relative to the other binary ASDs at all RH conditions studied, in particular at 94% RH. This is consistent with the moisture sorption profiles which

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

9

Figure 9. PXRD of ASDs stored at (a) 80 C/0% RH, (b) RT/94% RH, and (c) 40 C/75% RH.

indicate that PVP dispersions are the most hygroscopic, showing a large increase in moisture sorption, in particular above 75% RH. Furthermore, the IR spectra and the DSC data indicate that the PVP dispersions undergo AAPS on exposure to high RH. The rapid crystallization of the PVP-containing dispersions when stored at Table 1 Tg Values of Fresh CEX ASDs (6:4) and After Exposure to Different Relative Humidities for 3 Days ASD (6:4)

Unexposed ( C)

56% RH ( C)

75% RH ( C)

94% RH ( C)

CEX:PVP-K29/32 CEX:PVP-K12 CEX:PVP/VA CEX:HPMCAS CEX:HPMC

103 ± 2 84.4 ± 0.5 81 ± 2 64 ± 1.0 71 ± 1

64 ± 2 54.9 ± 0.1 56.3 ± 0.9 46.5 ± 0.5 53 ± 2

61 ± 2 47.4 ± 0.8 51.9 ± 0.7 43.8 ± 0.6 46.7 ± 0.6

37.4 ± 0.2 22 ± 3 42.1 ± 0.5 39 ± 1 38 ± 1

highly humid conditions can therefore most likely be accounted for by the increased hygroscopicity which in turn leads to disruption of drugepolymer interactions and subsequent crystallization. As compared with PVP dispersions, PVP/VA dispersions absorb much less water and hence are more resistant to both plasticization and AAPS. Similarly, Rumondor and Taylor17 found that for both felodipine and quinidine, the strength of hydrogen bonding of the drugs with PVP/VA was comparable to that with PVP, but the PVP/VA-containing mixture experienced a smaller extent of moisture-induced phase separation than the PVP-containing mixtures, which was attributed to reduced water uptake. In contrast, moisture-induced AAPS was absent in both indomethacin:PVP38 and indomethacin:PVP/VA systems. Interestingly, while HPMCAS dispersions of CEX absorb less water than PVP/VA dispersion, CEX:PVP/VA still outperforms CEX:HPMCAS at 94%

10

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11

stabilization. However, the release of the drug was relatively fast. The ternary ASD of itraconazole exhibited both fast dissolution and improved physical stability relative to the corresponding binary ASDs. Conclusions

Figure 10. Tg reduction in CEX ASDs (6:4) after exposure to different relative humidities for 3 days.

RH, most likely due to the stronger drugepolymer interaction strength as inferred from both the IR spectra and Tg values. Following the same argument, CEX:PVP/VA dispersions were much more stable than CEX:HPMC despite a similar moisture sorption profile. Finally, while replacing a small amount of the more hygroscopic PVP with a less hygroscopic cellulose derivative did result in reduced water uptake in ternary ASDs relative to the corresponding binary PVP ASD, this reduction was not large enough to result in a discernable difference in the physical stability. Thus, ternary systems were stable under dry storage conditions but crystallized after exposure to high RH. This is interesting, since the cellulose derivatives are extremely effective crystallization inhibitors in solution, even when present at extremely low concentrations. Clearly, this does not translate to the ASD matrix exposed to high RH when a hygroscopic polymer (PVP) is present. While the ASD with PVP absorbs more water at all relative humidities than ASDs with HPMCAS, this difference is less pronounced at or below 75% RH. The results shown in Figure 9c highlight the important interplay between temperature and RH effects and suggest that at 40 C/75% RH, drugepolymer interactions became the dominant factor in determining stability of the CEX dispersions. A comparison of the relative stability of the different ASDs at various storage conditions is summarized in Table S1. This study revealed that PVP/VA was very effective in stabilizing amorphous CEX at elevated temperature and humidity. We also found that PVP/VA was a good solution crystallization inhibitor for CEX. The nucleation induction time of CEX in the presence of 5-mg/mL PVP/VA is approximately 13 h (Fig. S11) as compared with 5 min in the absence of any polymer and <2 h in the presence of PVP under the same experimental conditions.34 Therefore, it appears that instead of using 2 different polymers, the same goal might be reached by using a single polymer: PVP/VA, which as a copolymer contains multiple functional groups. Nevertheless, the dissolution rate of the CEX:PVP/VA ASD was found to be much slower than that of the CEX:PVP:HPMCAS ternary ASDs32 whereby a lower level of supersaturation was generated under the same experimental conditions34 (Fig. S12). This may be undesirable for drugs such as CEX which is mainly used for pain management.31,45 Similarly, Six et al.29 found that itraconazole was miscible with PVP/VA, and the dispersion was stable against crystallization. But, the dissolution rate was slow, with only 45% drug release after 3 h. In contrast, Eudragit E100 was miscible with itraconazole only up to 13% w/w drug loading and failed to provide adequate physical

Polymers that can form stronger hydrogen bonds and increase Tg were found to be more effective at inhibiting crystallization of CEX from ASDs in the absence of moisture. In contrast, when subjected to high RH storage conditions, ASDs of CEX with stronger drugepolymer interaction and a lower hygroscopicity were found to be less susceptible to moisture-induced AAPS and crystallization, while the Tg of the dispersions was not predictive of the crystallization tendency. This study also demonstrated proof of concept for the approach of using miscible binary polymers to maximize crystallization inhibition in both the solid and solution phases by demonstrating that the addition of a second polymer did not substantially alter the crystallization kinetics. Acknowledgments The Dane O. Kildsig Center for Pharmaceutical Processing Research is acknowledged for providing funding for this project. References 1. Kwong AD, Kauffman RS, Hurter P, Mueller P. Discovery and development of telaprevir: an NS3-4A protease inhibitor for treating genotype 1 chronic hepatitis C virus. Nat Biotechnol. 2011;29(11):993-1003. 2. Bollag G, Hirth P, Tsai J, et al. Clinical efficacy of a RAF inhibitor needs broad target blockade in BRAF-mutant melanoma. Nature. 2010;467(7315):596-599. 3. Van den Mooter G, Wuyts M, Blaton N, et al. Physical stabilisation of amorphous ketoconazole in solid dispersions with polyvinylpyrrolidone K25. Eur J Pharm Sci. 2001;12(3):261-269. 4. Khougaz K, Clas S-D. Crystallization inhibition in solid dispersions of MK-0591 and poly(vinylpyrrolidone) polymers. J Pharm Sci. 2000;89(10):1325-1334. 5. Matsumoto T, Zografi G. Physical properties of solid molecular dispersions of indomethacin with poly(vinylpyrrolidone) and poly(vinylpyrrolidone-covinylacetate) in relation to indomethacin crystallization. Pharm Res. 1999;16(11):1722-1728. 6. Marsac P, Konno H, Taylor LS. A comparison of the physical stability of amorphous felodipine and nifedipine systems. Pharm Res. 2006;23(10):2306-2316. 7. Miyazaki T, Yoshioka S, Aso Y, Kojima S. Ability of polyvinylpyrrolidone and polyacrylic acid to inhibit the crystallization of amorphous acetaminophen. J Pharm Sci. 2004;93(11):2710-2717. 8. Doherty C, York P. Evidence for solid-and liquid-state interactions in a furosemide-polyvinylpyrrolidone solid dispersion. J Pharm Sci. 1987;76(9): 731-737. n ML, Manzo RH, Olivera ME. Eudragit E100 as a drug carrier: the 9. Guzma remarkable affinity of phosphate ester for dimethylamine. Mol Pharm. 2012;9(9):2424-2433. 10. Quinteros DA, Rigo VR, Kairuz AFJ, Olivera ME, Manzo RH, Allemandi DA. Interaction between a cationic polymethacrylate (Eudragit E100) and anionic drugs. Eur J Pharm Sci. 2008;33(1):72-79. 11. Wegiel LA, Mauer LJ, Edgar KJ, Taylor LS. Mid-infrared spectroscopy as a polymer selection tool for formulating amorphous solid dispersions. J Pharm Pharmacol. 2014;66(2):244-255. 12. Zografi G. States of water associated with solids. Drug Dev Ind Pharm. 1988;14(14):1905-1926. 13. Katkov II , Levine F. Prediction of the glass transition temperature of water solutions: comparison of different models. Cryobiology. 2004;49(1):62-82. 14. Hancock BC, Zografi G. The relationship between the glass transition temperature and the water content of amorphous pharmaceutical solids. Pharm Res. 1994;11(4):471-477. 15. Andronis V, Yoshioka M, Zografi G. Effects of sorbed water on the crystallization of indomethacin from the amorphous state. J Pharm Sci. 1997;86(3):346-351. 16. Marsac PJ, Rumondor AC, Nivens DE, Kestur US, Stanciu L, Taylor LS. Effect of temperature and moisture on the miscibility of amorphous dispersions of felodipine and poly (vinyl pyrrolidone). J Pharm Sci. 2010;99(1):169-185. 17. Rumondor ACF, Taylor LS. Effect of polymer hygroscopicity on the phase behavior of amorphous solid dispersions in the presence of moisture. Mol Pharm. 2010;7(2):477-490. 18. Purohit HS, Taylor LS. Phase separation kinetics in amorphous solid dispersions upon exposure to water. Mol Pharm. 2015;12(5):1623-1635.

T. Xie, L.S. Taylor / Journal of Pharmaceutical Sciences xxx (2016) 1-11 19. Williams HD, Trevaskis NL, Charman SA, et al. Strategies to address low drug solubility in discovery and development. Pharmacol Rev. 2013;65(1):315-499. 20. Zografi G, Hancock B. Proceedings of 53rd International Congress of Pharmaceutical Sciences, 1994, 405e419. 21. Hancock BC, Zografi G. Characteristics and significance of the amorphous state in pharmaceutical systems. J Pharm Sci. 1997;86(1):1-12. 22. Konno H, Handa T, Alonzo DE, Taylor LS. Effect of polymer type on the dissolution profile of amorphous solid dispersions containing felodipine. Eur J Pharm Biopharm. 2008;70(2):493-499. 23. Kestur US, Taylor LS. Role of polymer chemistry in influencing crystal growth rates from amorphous felodipine. Crystengcomm. 2010;12(8):2390-2397. 24. Konno H, Taylor LS. Influence of different polymers on the crystallization tendency of molecularly dispersed amorphous felodipine. J Pharm Sci. 2006;95(12):2692-2705. 25. Rumondor ACF, Stanford LA, Taylor LS. Effects of polymer type and storage relative humidity on the kinetics of felodipine crystallization from amorphous solid dispersions. Pharm Res. 2009;26(12):2599-2606. 26. Trasi NS, Taylor LS. Effect of polymers on nucleation and crystal growth of amorphous acetaminophen. Crystengcomm. 2012;14(16):5188-5197. 27. Trasi NS, Oucherif KA, Litster JD, Taylor LS. Evaluating the influence of polymers on nucleation and growth in supersaturated solutions of acetaminophen. Crystengcomm. 2015;17(6):1242-1248. 28. Sakurai A, Sakai T, Sako K, Maitani Y. Polymer combination increased both physical stability and oral absorption of solid dispersions containing a low glass transition temperature drug: physicochemical characterization and in vivo study. Chem Pharm Bull. 2012;60(4):459-464. 29. Six K, Verreck G, Peeters J, Brewster M, Van den Mooter G. Increased physical stability and improved dissolution properties of itraconazole, a class II drug, by solid dispersions that combine fast- and slow-dissolving polymers. J Pharm Sci. 2004;93(1):124-131. 30. Paudel A, Worku ZA, Meeus J, Guns S, Van den Mooter G. Manufacturing of solid dispersions of poorly water soluble drugs by spray drying: formulation and process considerations. Int J Pharm. 2013;453(1):253-284. 31. Paulson SK, Vaughn MB, Jessen SM, et al. Pharmacokinetics of celecoxib after oral administration in dogs and humans: effect of food and site of absorption. J Pharmacol Exp Ther. 2001;297(2):638-645. 32. Xie T, Taylor LS. Dissolution performance of high drug loading celecoxib amorphous solid dispersions formulated with polymer combinations. Pharm Res. 2016;33(3):739-750.

11

33. Kaushal AM, Chakraborti AK, Bansal AK. Ftir studies on differential intermolecular association in crystalline and amorphous states of structurally related non-steroidal anti-inflammatory drugs. Mol Pharm. 2008;5(6):937-945. 34. Nakamoto K, Margoshes M, Rundle RE. Stretching frequencies as a function of distances in hydrogen bonds. J Am Chem Soc. 1955;77(24):6480-6486. 35. Van Eerdenbrugh B, Taylor LS. An ab initiopolymer selection methodology to prevent crystallization in amorphous solid dispersions by application of crystal engineering principles. Crystengcomm. 2011;13(20): 6171-6178. 36. Coleman MM, Painter PC, Graf JF. Specific Interactions and the Miscibility of Polymer Blends. Pensilvania: Technomic Publishing Company, Inc; 1995: 221-304. 37. Marks JA, Wegiel LA, Taylor LS, Edgar KJ. Pairwise polymer blends for oral drug delivery. J Pharm Sci. 2014;103(9):2871-2883. 38. Rumondor AC, Marsac PJ, Stanford LA, Taylor LS. Phase behavior of poly (vinylpyrrolidone) containing amorphous solid dispersions in the presence of moisture. Mol Pharm. 2009;6(5):1492-1505. 39. Taylor LS, Langkilde FW, Zografi G. Fourier transform Raman spectroscopic study of the interaction of water vapor with amorphous polymers. J Pharm Sci. 2001;90(7):888-901. 40. Qian F, Huang J, Zhu Q, et al. Is a distinctive single Tg a reliable indicator for the homogeneity of amorphous solid dispersion? Int J Pharm. 2010;395(1): 232-235. 41. Qian F, Huang J, Hussain MA. Drugepolymer solubility and miscibility: stability consideration and practical challenges in amorphous solid dispersion development. J Pharm Sci. 2010;99(7):2941-2947. 42. Lu Q, Zografi G. Phase behavior of binary and ternary amorphous mixtures containing indomethacin, citric acid, and PVP. Pharm Res. 1998;15(8): 1202-1206. 43. Marsac P, Shamblin S, Taylor LS. Theoretical and practical approaches for prediction of drugepolymer miscibility and solubility. Pharm Res. 2006;23(10): 2417-2426. 44. Bhugra C, Pikal MJ. Role of thermodynamic, molecular, and kinetic factors in crystallization from the amorphous state. J Pharm Sci. 2008;97(4):1329-1349. 45. Terebetski JL, Cummings JJ, Fauty SE, Michnik-Kohn B. Combined use of crystalline sodium salt and polymeric precipitation inhibitors to improve pharmacokinetic profile of ibuprofen through supersaturation. AAPS PharmSciTech. 2014;15(5):1334-1344.