RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Effect of Different “States” of Sorbed Water on Amorphous Celecoxib GANESH SHETE,1 SWATHI KUNCHAM,2 VIBHA PURI,3 RAHUL P. GANGWAL,4 ABHAY T. SANGAMWAR,4 ARVIND KUMAR BANSAL1 1
Department of Pharmaceutics, National Institute of Pharmaceutical Education and Research (NIPER), S.A.S. Nagar, Punjab 160062, India Generics FR&D, Dr. Reddy’s Laboratories, Bachupally, Hyderabad 500072, India 3 Novartis-MIT Centre for continuous manufacturing, Cambridge, Massachusetts 02139 4 Department of Pharmacoinformatics, National Institute of Pharmaceutical Education and Research (NIPER), S.A.S. Nagar, Punjab 160062, India 2
Received 13 November 2013; revised 7 April 2014; accepted 14 April 2014 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23999 ABSTRACT: Glass transition temperature (Tg ) of an amorphous drug is a vital physical phenomenon that influences its visco-elastic properties, physical, and chemical stability. Water acts as a plasticizer for amorphous drugs thus increasing their recrystallization kinetics. This reduces the solubility advantage of an amorphous drug. Hence, there is an interest in understanding the relationship between water content and Tg of amorphous drug. We have studied the effect of “state” of sorbed water on Tg of amorphous celecoxib (ACLB). ACLB was allowed to sorb water at relative humidity of 33%, 53%, 75%, and 93%. ALCB showed biphasic sorption of water designated as “bound” and “solvent-like” state of water associated with ACLB. Molecular modeling studies provided deeper insights into the interaction of water with ACLB. A distinct co-relationship between the state of water and its plasticization capacity was observed. Bound state of water had a very profound effect on the fall in experimentally observed Tg (Tg-exp ) value. Solvent-like state of water had little impact on Tg-exp value. Tg of ACLB–water mixture was predicted by Gordon–Taylor equation (Tg-pre ). The deviations in Tg-exp and Tg-pre were correlated to volume C 2014 non-additivity and non-ideal mixing. This study has implications on the development of formulations based on amorphous forms. Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci Keywords: Amorphous; glass transition; water sorption; water in solids; molecular modeling
INTRODUCTION Amorphous solids offer several advantages in pharmaceuticals; however, their industrial application has remained limited because of their inherent tendency to convert to the stable crystalline state.1 Amorphous materials have higher solubility than their crystalline counterparts because of the disordered state and higher free energy. Effective “stabilization” of amorphous state can enable exploitation of their solubility advantage.2 Interaction of amorphous pharmaceuticals with water has always evoked interest. Water is ubiquitously present in pharmaceutical systems and more so in amorphous state because of their higher hygroscopicity. Water is a strong plasticizer with a glass transition temperature (Tg ) of −138◦ C.3 It has a deleterious effect on the amorphous substances as it lowers their Tg and accelerates the recrystallization process. Latter reduces the solubility advantage achievable from the amorphous systems.4 The impact of water on various properties of amorphous systems like Tg and recrystallization behavior has been studied on pharmaceuticals like sequinavir, lamotrigine mesylate, nifedipine, felodipine, sucrose, indomethacin, and lactose3,5–16 using numerous methods. However, reports investigating the effect of “state” of sorbed water on Tg are lacking in literature.
Correspondence to: Arvind Kumar 2214682ext2126; Fax: +91-172-2214692; [email protected])
Bansal (Telephone: +91-172E-mail: [email protected];
Journal of Pharmaceutical Sciences C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association
A review paper published in 1988 classified water sorption by amorphous solids into two different types: “bound state” and “solvent-like” state of water.17 “Bound water” involves specific hydration directly to the molecules of amorphous solid and affects its physical and chemical phenomena. At higher water contents, clusters are formed and a “solvent-like” state of water is generated. Because this water is not available directly in the vicinity of amorphous solid, it exhibits limited effects on the latter. The concepts proposed in the aforementioned reference had not been justified by experimental findings/examples.17 In contrast to the two-state model, Hancock et al.18 had suggested a “mutual miscibility” or “plasticizer” model to explain the water vapor sorption by amorphous material. This model envisages a continuum of chemical, physical, and energetic states.18 The study was based on hydrophilic polymers like polyvinylpyrrolidone (PVP) that allows significant interaction with water molecules. The Tg for drug–water binary system can be predicted using theoretical approaches like Fox equation or Gordon–Taylor equation. However, deviations of experimentally observed Tg (Tg-exp ) from theoretically predicted Tg (Tg-pre ) values are common. These deviations have been attributed to various reasons like molecular mobility,14 differences in molecular weight,19 free volume,5 and/or molecular size.3 The purpose of the present work is to investigate the effect of “state” of sorbed water on Tg-exp of amorphous celecoxib (ACLB). To the best of our knowledge, this is the first report on the impact of “state” of sorbed water on Tg-exp of an amorphous form of a low-molecular-weight pharmaceutical molecule. Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
1
2
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Powder X-Ray Diffraction Powder X-ray diffraction (PXRD) patterns of samples were recorded at room temperature on Bruker’s D8 Advance diffractometer (Karlsruhe, West Germany) with Cu K" radiation ˚ at 40 kV and 40 mA passing through nickel filter (1.54 A), with divergence slit (0.5◦ ), antiscattering slit (0.5◦ ), and receiving slit (0.1 mm). PXRD scans were performed with a step size of 0.01◦ and step time of 1 s over an angular range of 3◦ –40◦ 2θ. Obtained diffractograms were analyzed with DIFFRACplus EVA (version 9.0) diffraction software. Microscopy Figure 1. Chemical structure of celecoxib.
Further, comprehensive reasons for variability in Tg-exp and Tg-pre have been discussed, considering contributors like volume non-additivity and non-ideality in the mixing of water and ACLB. This understanding has significant implications on drug development using amorphous system.
MATERIALS AND METHODS Material The crystalline form of celecoxib (CLB) was purchased from Unichem Laboratories Ltd. (Raigad, India). The purity of the sample was confirmed to be greater than 99.0% by the certificate of analysis provided by Unichem Laboratories. CLB is a Biopharmaceutics Classification System class II drug.20 ACLB has been reported to have a Tg-exp of 51◦ C–57 ◦ C and mean molecular relaxation time of 2.79 h, as calculated by Kohlrausch–Williams–Watts equation (β = 0.274).21 Chemical structure of CLB is shown in Figure 1. Methods Preparation of ACLB In situ ACLB was generated using differential scanning calorimetry (DSC). Accurately weighed amount of CLB (about 5 mg) was taken in an aluminum pan and heated to 185◦ C at heating rate of 20◦ C/min, held isothermally for 1 min to ensure complete melting followed by quench cooling to 25◦ C at cooling rate of 20◦ C/min under dry nitrogen purge of 50 mL/min. HPLC analysis indicated that no chemical degradation occurred during the generation of ACLB. Ex situ ACLB was prepared in comparatively larger quantity to carry out water sorption studies. CLB was melted in R ; Associated a stainless steel beaker on a hot plate (Sonar Scientific Technologies, New Delhi, India), and subsequently quench cooled over crushed ice. The quench-cooled product was passed through British Standard Sieve 44 resulting in a particle size range of about 5–50 :m with a D90 of 40 :m. No devitrification occurred during this processing. HPLC analysis indicated that no chemical degradation occurred during the generation of ACLB. All operations required for ex situ ACLB generation were performed in a controlled humidity condition [< 15% relative humidity (RH)] and at ambient temperature. Initial water content of ACLB samples was found to lesser than 0.05% (w/w), as determined by Karl Fisher titrimetry (716 DMS Titrino; Metrohm Limited, Herisau, Switzerland). Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
Samples were visually observed under Leica DMLP polarized microscope (Leica Microsystems Wetzlar GmbH, Wetzlar, Germany) in a bright light and cross-polarized light mode. Photomicrographs were acquired using Leica DC 300 camera and analyzed using Linksys 32 software. Water Vapor Sorption For the assessment of the effect of sorbed water on Tg-exp , ACLB samples were exposed to 33%, 53%, 75%, and 93% RH in an open DSC pan at 25◦ C in a vacuum desiccator up to 5 h (h). The samples were withdrawn at different time points up to 5 h. Water sorption of ACLB was measured gravimetrically from the difference between the weight of the dry sample immediately after its preparation and that after humidification. All the weights were measured using Sartorius GC 2502 weighing balance (New York, USA). These samples were subsequently analyzed in DSC after hermetically sealing the DSC pans. All the measurements were carried out in triplicates. Differential Scanning Calorimetry Conventional DSC experiments were conducted using DSC Q2000 (TA Instruments, Delaware, USA) equipped with a refrigerated cooling system and operating with Universal Analysis 2000 software version 4.5A. The instrument was calibrated for heat flow and temperature with high purity standards of Indium and Zinc. An important concern in the study of the effect of water on amorphous drug is the determination of Tg-exp in DSC. Several reports are available in the literature regarding the use of DSCbased protocols like (1) a single heating run of 0◦ C–200◦ C and (2) two heating runs, wherein first heating run is carried out up to a temperature limit of Tg + 20◦ C to eliminate the thermal history of sample and second heating run is used to determine Tg-exp .11 Tg-exp values determined from either of the protocols did not show any significant differences in our study. Hence, a single heating run protocol was adopted. Samples were scanned at a heating rate of 20◦ C/min in hermetically sealed aluminum pans from 0◦ C to 200◦ C with nitrogen purging (50 mL/min). Tg-exp , Tc , and Tm have been reported as onset temperature values determined in first heating run. Water affects the Tg-exp of material only when it is in physical contact with the amorphous matrix. For this purpose, hermetically sealed DSC pans were used for the analysis. About 7– 10 mg sample was added into DSC pan to keep headspace minimum during sealing of pan and lid to avoid evaporation of water into headspace during heating. All the studies have been performed in triplicates and average values are reported. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
3
Figure 2. PXRD patterns of (i) CLB and (ii) ACLB generated by in situ and ex situ method. Inset depicts zoomed versions of PXRD patterns of in situ and ex situ ACLB.
Molecular Modeling All the molecular modeling calculations were performed using SYBYL 7.1 package installed on a Silicon Graphics Fuel Workstation running IRIX 6.5. The CLB starting ensemble was built from available crystal data. Random arrangement of 55 CLB molecules, simulating amorphous form, was modeled by molecular silverware method. The molecules were later minimized by applying molecular mechanics force field (MMFF94) with the conjugate gradient method. The minimization was terminated when the energy gradient convergence criterion of 0.05 kcal/mol was reached or when the 100 step minimization cycle was exceeded. The optimized model was further subjected to molecular dynamics (MD) simulation using simulated annealing approach. The system was heated to 185◦ C to transform the crystal into the liquid state, and equilibrated for 1000 femtoseconds (fs) at 185◦ C, and finally cooled to 25◦ C. The technique was analogous to experimental procedure adopted for preparing ACLB by melting the drug and suddenly cooling it. Molecular dynamics simulation was performed for simulation of water vapor sorption by ACLB. The lattice of 10 × ˚ 3 was created, with 0.0279 g/cm3 density of water 10 × 10 A vapor at 93% RH and at 25◦ C, by molecular silverware method. The constructed system was minimized by applying molecular mechanics force field (MMFF94) with the 100 minimization cycle of steepest decent method and 100 minimization cycle of DOI 10.1002/jps.23999
conjugate gradient method. Periodic boundary conditions were imposed in order to eliminate the surface effects. The systems were equilibrated using constant number of atoms, volume, and energy ensemble for 50 picoseconds (ps) with a 0.5 fs time step. Then, MD simulation was performed on minimized equilibrated water–ACLB system at 25◦ C under constant moles, volume, and temperature (NVT) condition. The number of steps of NVT MD simulation was 500 ps with time step of 1 fs and output frequency was after every 100 steps.
RESULTS Solid-State Characterization Figure 2 shows PXRD scans of CLB and ACLB generated by in situ and ex situ method. PXRD analysis of CLB revealed its crystalline nature. CLB exhibited peaks at 2θ values of 5.36◦ , 10.72◦ , 13.01◦ , 14.80◦ , 16.10◦ , 19.65◦ , 21.52◦ , 23.45◦ , 25.38◦ , 27.00◦ , and 29.59◦ corresponding to crystalline form III. ACLB generated by in situ and ex situ method exhibited halo pattern in PXRD at room temperature. DSC scans of CLB, in situ ACLB and ex situ ACLB are shown in Figure 3. CLB showed a single sharp endotherm corresponding to melting (Tm ) at 162.3◦ C. In situ generated ACLB exhibited a Tg-exp of 57.6◦ C. No recrystallization or melting event was Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
4
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 3. DSC traces of (i) CLB and (ii) ACLB generated by in situ and ex situ method.
evident in this sample. However, it was not possible to generate in situ sample in quantities sufficient for water sorption studies. Hence, ACLB was generated using an ex situ protocol. Ex situ generated ACLB exhibited a halo pattern in PXRD (Fig. 2) and absence of birefringence in polarized light microscopy studies (data not shown), thus establishing its amorphous nature. In DSC, ex situ ACLB exhibited a Tg-exp at 53.8◦ C and heat induced recrystallization (Tc ) at 126.1◦ C–152.6◦ C followed by a broad melting endotherm at 161.4◦ C. The recrystallization can be attributed to nonisothermal run in DSC. The ex situ ACLB was used for subsequent water sorption studies.
Water Sorption Behavior of ACLB Exposure of ACLB to 33%, 53%, 75%, and 93% RH conditions at 25◦ C in desiccators led to recrystallization of ACLB and the recrystallization rate increased from 33% to 93% RH condition, as was evident by birefringence observed under cross polarized light in microscope (data not shown). The minimum time required for onset of recrystallization at 93% RH (at which the recrystallization rate was maximum) was approximately 5 h. The presence of crystallinity in amorphous material is likely to change the response of amorphous phase toward water in terms of plasticization and recrystallization. To avoid these possibilities, all the water sorption studies at 33%, 53%, 75%, and 93% RH were performed only up to 5 h wherein the sample was completely in amorphous state. ACLB sorbed water with increasing rates at 33%, 53%, 75%, and 93% RH linearly. ACLB sorbed maximum water of 1.88% (w/w) at 93% RH and at 5 h. ACLB was found to be nonhygroscopic as per Callahan hygroscopicity classification.22 Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
Effect of Sorbed Water on Tg Nonisothermal heating run of ACLB with sorbed water (ACLB– water mixture) in DSC demonstrated phenomena of Tg-exp , Tc , and Tm . Table 1 lists Tg-exp , Tc , and Tm values for ACLB–water mixture. The theoretically expected value of Tg (Tg-pre ) was calculated by using Gordon–Taylor equation, which is given as follows3 : Tg−pre =
w1 Tg1 + kw2 Tg2 w1 + kw2
(1)
where Tg1 and Tg2 are Tg values of water and ACLB, respectively. w represents the weight fraction and k is ratio of free volumes given by, k=
1 Tg1 2 Tg2
(2)
where ρ 1 and ρ 2 indicate the true densities of water and ACLB, respectively. Figure 4 shows the behavior of Tg-exp and Tg-pre with respect to sorbed water. The Tg-exp showed a fall with increasing water content because of the plasticizing effect of water. Interestingly, Tg-exp demonstrated steepest fall till water content of 1% (w/w). Thereafter, a little effect of water on Tg-exp was observed. Reduction in Tc values from 126◦ C to 76◦ C was observed with increasing water content. DSC analysis was performed on samples that had not shown recrystallization during storage and thus residual crystallinity did not contribute to Tc . Thus, the observed reduction in Tc was contributed only by the water in the amorphous matrix.19 DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Table 1.
Tg-exp , Tg-pre , Tc , and Tm Values for ACLB–Water Mixture
Water Content (%, w/w) Tg-exp (◦ C)a Tg-pre (◦ C)b Tc (◦ C)a Tm (◦ C)a 0 0.17 0.20 0.23 0.24 0.33 0.39 0.42 0.46 0.54 0.60 0.66 0.68 0.74 0.79 0.85 0.90 0.95 1.00 1.13 1.27 1.40 1.88 a b
53.8 52.4 51.3 50.6 47.5 46.8 45.1 44.5 43.2 42.5 41.9 41.8 41.7 40.8 39.1 37.1 36.2 35.1 34.9 34.5 34.2 34.1 31.0
53.8 53.8 51.3 51.2 51.0 50.9 50.4 50.1 49.9 49.6 49.2 48.8 48.5 48.4 48.0 47.7 47.4 47.2 46.9 46.6 45.9 45.1 44.3
126.1 108.2 105.8 103.0 103.9 102.0 101.8 101.4 99.8 98.5 95.4 90.4 89.6 88.0 87.5 85.6 84.3 83.8 82.8 76.9 76.5 76.8 76.9
161.4 160.5 160.3 160.6 160.2 161.7 160.4 160.9 160.2 160.1 160.3 161.8 160.4 160.5 161.4 160.3 160.1 160.3 161.8 160.4 160.5 160.5 160.7
Mean of experimentally observed values for n=3. Predicted values by Gordon–Taylor equation.
DISCUSSION Mechanism of Water Sorption by ACLB Matrix The sorption of water up to 1.00% (w/w) resulted in drop of Tg-exp values, thus pointing toward close interaction of water with ACLB. However, beyond 1% (w/w) water content, no appreciable change in Tg-exp was observed. Our findings do not support concept of “continuum model” of water sorption proposed by
5
Hancock et al.18 The observations and proposal for continuum model was based on the experiments carried out with PVP, which is a hydrophilic polymer. ACLB possesses relatively hydrophobic molecular environment (discussed in detail in the preceding section) and exhibited a behavior that closely resembled with concept proposed by Prof. Zografi.17 Hence, in the present work, the sorption of water up to 1.00% (w/w) can be designated as “bound” state of water. The “bound” water is in closest vicinity of solids and directly hydrates the molecules of solid matrix. This type of water affects the molecular level properties of solids as well as chemical reactions. The second phase of water sorption by ACLB from 1% to 1.88% (w/w) can be designated as the “solvent-like” state of water, which was found to have negligible effect on Tg-exp . At higher water content, water molecules form clusters and give rise to “solventlike” state.17 These types of clusters cannot effectively contact the hydrophobic amorphous matrix and thus exert negligible effects on amorphous material. The molecular environment possessed by the pharmaceutical material determines its interaction with water. In case of a low-molecular-weight hydrophobic organic pharmaceutical compound like ACLB, water can interact only up to a limited extent, as governed by the hydrophilic sites in the molecular structure. A limited interaction is possible between molecules of water and ACLB, after which continuum is disturbed and clusters of water molecules are formed over amorphous bed.18 Further, CLB is a hydrophobic compound and it is very less likely that it will show capillary condensation. The “mutual miscibility” or “continuum” model suggested by Hancock et al.18 thus seems to be inapplicable in the case of ACLB–water system. Molecular Understanding of Interaction of Water with ACLB In the last few decades, molecular modeling has been utilized in rationalizing the experimental observations and provides deeper insights into “difficult-to-explain” events. We carried
Figure 4. Comparison of fall in Tg-exp () and Tg-pre (О) with increasing water content for ACLB–water mixture. DOI 10.1002/jps.23999
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
6
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 5. (a) Computer simulation of molecular arrangement of ACLB, (b) interaction of water molecules with ACLB bed, (c) pictorial depiction of states of water associated with ACLB matrix, and (d) hydrogen bonding interactions between water and CLB molecules; based on the special constraints, all CLB molecules do not show all types of interactions.
out the MD study to understand the moisture sorption behavior by ACLB. Figure 5a shows ACLB matrix simulated on the computer and Figure 5b shows interactions of ACLB with water. Water was preferentially found to interact with hydrophilic sites of ACLB through H-bonding. This type of sorbed water can be designated as “bound” type of water. After all the hydrophilic sites are saturated, the water molecules started to form clusters around already sorbed water that can be termed as “solvent-like” state of water. Figure 5c depicts bound and solvent-like states of water associated with ACLB matrix. Molecular lipophilic surface potential (MLSP) is a measure of lipophilicity of a molecule, which describes combined lipophilic influence of all fragments contained in it.23 Evaluation of MLSP of crystalline CLB revealed contribution of –SO2 NH2 group toward hydrophilicity and contribution of lipophilic methyl phenyl moiety toward lipophilicity.24 Findings of molecular modeling studies provided interesting insights. Hydrogen bonding (H-bonding) was observed between –SO2 NH2 group of CLB and water molecules. Some of CLB molecules also showed additional H-bond interaction with water through –NH group of Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
pyrazole ring. This H-bonding between ACLB and bound water affects the solid-state properties of the former like Tg and recrystallization tendency. The H-bonding interactions between water and CLB molecules are shown in Figure 5d. Evidence of H-bonds between ACLB and water has also been established earlier from differential heat of sorption at 25◦ C/54% RH (155.04 ± 5.02 kJ/mol), which was found to be greater than the heat of condensation of water (44 kJ/mol).25 We were able to unveil the average interaction occurring between single CLB and water molecules (Fig. 5d). However, not all CLB molecules in amorphous phase participated in similar interactions with water molecules. Some of the domains of CLB molecules involved in interactions with water are “preengaged” because of their interaction with neighboring CLB molecules. Amorphous phase lacks long-range order, but possesses short-range order. This short-range order presents differential molecular environment around various CLB molecules. Although from a molecular perspective, each CLB molecule offers possibility of interacting with five water molecules, but the supramolecular environment of amorphous phase forces different CLB molecules to interact with water in a different manner. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
This prevented us from quantifying the average number of water molecules that interact with single CLB molecules. Experimental evidence for limited interaction of water with ACLB has been reported in our previous publications.25 A significant difference between solubility parameter (δ t ) values of CLB (21.01) and water (47.9) suggested their immiscibility and thus limited molecular interaction. δ t of CLB was calculated by the group contribution method and found to be 21.01,26,27 whereas δ t of water was taken from literature as 47.9.28 Mixing of Water with ACLB The plasticizing effect of water on amorphous phase of drugs can be predicted by calculating Tg-pre of a drug–water mixture using Gordon–Taylor equation. Tg-exp values were found be lower than that of Tg-pre . The deviation between Tg-exp and Tg-pre became prominent with the increase in the water content up to 1.0% (w/w). Thereafter, the deviation narrowed down at sorbed values above 1% (w/w), as shown in Figure 4. The differences observed between Tg-exp and Tg-pre can be explained by the thermodynamic calculations and discussion provided below. The thermodynamic calculations were meant to further explore the mixing behavior of water with ACLB. Gordon–Taylor equation is based on the free volume theory and assumes ideal mixing. It describes the behavior of a “mis-
7
cible” amorphous mixture of a binary system. However, most of the amorphous molecular dispersions are supersaturated “solutions,” as during formation, materials are trapped in a glass, with no measurable phase separation of individual phases. In DSC measurements, miscibility is evidenced by a single Tg .29 The constant k of the equation can be calculated from known parameters and allows use of weight fraction to describe compositions in such systems. Ideal behavior of mixtures can be predicted in terms of the enthalpy change of mixing which is zero for ideal mixing, and negative or positive for non-ideal mixing. The deviation between Tg-exp and Tg-pre indicated nonadditivity of volume because of the non-uniform distribution of the free volume between ACLB and water, and huge difference in their Tg values. These deviations are a result of molecular interactions between ACLB and water. As evidenced by molecular modeling studies and reports from literature,25 ACLB and water can form H-bonds. However, a weaker set of inter-molecular interactions between CLB and water as compared with CLB– CLB + water–water leads to volume expansion and volume non-additivity.30 The phenomenon of volume non-additivity is usually associated with the non-ideality of mixing. Non-ideality of mixing can be predicted by determination of excess enthalpy (HE ), entropy (SE ), and free energy (GE ) of mixing.30 These
Figure 6. Enthalpy, entropy, and free energy of mixing for ACLB–water mixture. DOI 10.1002/jps.23999
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
8
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
parameters were calculated with respect to % (w/w) water content and are shown in Figure 6. Average Tg-exp values were used for thermodynamic calculations. Cp values at Tg for water and ACLB were used from literature as 0.088931 and 0.4421 J/K32 , respectively. The net enthalpy of mixing increased with the increase in water content and indicated disruption of H-bonds, thus demanding more energy for interaction. Simultaneously, the breakage of “native order” of disordered amorphous matrix of ACLB by the addition of water increased the net entropy of mixing.33 Further, free energy represents the difference in the free energy of the mixture at Tg-exp and free energy of the mixture at temperature T assuming that the entropy of mixing is purely combinatorial and the mixing is non-thermal. Free energy decreased because of the relatively higher increase in the factor of Tg-exp as compared with the HE . The evaluation of enthalpy, entropy, and free energy of mixing indicated non-ideal mixing behavior of ACLB–water system. Biphasic curves were observed with leveling off in the curves after 1% (w/w) sorbed water content, indicating that the nonideality of mixing had attained its peak value at 1% (w/w) sorbed water. This further supports the fact that water is in direct contact with ACLB matrix only up to 1% (w/w) and represents “bound” state of water. The leveling of the curves after 1% (w/w) indicates that the non-interactive nature of water represents its “solvent-like” state.
CONCLUSIONS The interaction of water with pharmaceutical solids is critical during processing and shelf life stability. Water is ubiquitously present in drugs and excipients and can impact solid phase behavior of ingredients. Although total water content is important, the state in which water is present is even more critical. Water present in the “bound” state participates in molecular interactions most profoundly thus affecting the solid state behavior like Tg and recrystallization tendency. Critical assessment of the sorbate properties provides insights into its interaction with water. Volume non-additivity and non-ideal mixing provide the underlying thermodynamic basis for these interactions. Additional studies on materials with diverse physicochemical properties can generate guidelines for the role of water in performance of drug products, containing amorphous pharmaceuticals.
ACKNOWLEDGMENTS The authors thank Dr. Bruno Hancock and Prof. George Zografi for their valuable inputs and helpful discussions. Ganesh Shete and Rahul P. Gangwal acknowledge the financial support from Council of Scientific and Industrial Research, Government of India.
REFERENCES 1. Craig DQ, Royall PG, Kett VL, Hopton ML. 1999. The relevance of the amorphous state to pharmaceutical dosage forms: Glassy drugs and freeze dried systems. Int J Pharm 179(2):179–207. 2. Hancock BC, Parks M. 2000. What is the true solubility advantage for amorphous pharmaceuticals? Pharm Res 17:397–403. 3. Hancock BC, Zografi G. 1994. The relationship between the glass transition temperature and the water content of amorphous pharmaceutical solids. Pharm Res 11(4):471–477. Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
4. Buckton G, Patricia D. 1999. Assessment of disorder in crystalline powders-a review of analytical techniques and their application. Int J Pharm 179:141–158. 5. Andronis V, Yoshioka M, Zografi G. 1997. Effects of sorbed water on the crystallization of indomethacin from the amorphous state. J Pharm Sci 86:346–351. 6. Buckton G, Darcy P. 1996. Water mobility in amorphous lactose below and close to the glass transition temperature. Int J Pharm 136:141– 146. 7. Burnett DJ, Thielmann F, Sokoloski T, Brum J. 2006. Investigating the moisture-induced crystallization kinetics of spray-dried lactose. Int J Pharm 313(1–2):23–28. 8. Grigera JR, Viviana B. 2008. Molecular mobility and ageing of sugar glasses. Food Chem 106:1314–1317. 9. Hu X, Kaplan D, Cebe P. 2007. Effect of water on the thermal properties of silk fibroin. Thermochim Acta 461:137–144. 10. Imamura K, Atsushi F, Sakaura K, Sugita T, Sakiyama T. 2002. Water sorption and glass transition behaviors of freeze-dried sucrosedextran mixtures. J Pharm Sci 91:2175–2181. 11. Konno H, Taylor LS. 2008. Ability of different polymers to inhibit the crystallization of amorphous felodipine in the presence of moisture. Pharm Res 25(4):969–978. 12. Pittia P, Sacchetti G. 2008. Antiplasticization effect of water in amorphous foods. A review. Food Chem 106:1417–1427. 13. Royall PG, Craig DQ, Doherty C. 1999. Characterisation of moisture uptake effects on the glass transitional behaviour of an amorphous drug using modulated temperature DSC. Int J Pharm 192(1):39–46. 14. Schmitt E, Davis CW, Long ST. 1996. Moisture-dependent crystallization of amorphous lamotrigine mesylate. J Pharm Sci 85(11):1215– 1219. 15. Schubnell M, Schawe JE. 2001. Quantitative determination of the specific heat and the glass transition of moist samples by temperature modulated differential scanning calorimetry. Int J Pharm 217(1– 2):173–181. 16. Shamblin SL, Zografi G. 1999. The effects of absorbed water on the properties of amorphous mixtures containing sucrose. Pharm Res 16:1119–1124. 17. Zografi G. 1988. States of water associated with solids. Drug Dev Ind Pharm 14:1905–1926. 18. Hancock BC, Zografi G. 1993. The use of solution theories for predicting water vapor absorption by amorphous pharmaceutical solids: A test of the Flory–Huggins and Vrentas models. Pharm Res 10(9):1262– 1267. 19. Saleki-Gerhardt A, Zografi G. 1994. Non-isothermal and isothermal crystallization of sucrose from the amorphous state. Pharm Res 11:1166–1173. 20. Simon LS, Lanza FL, Lipsky PE, Hubbard RC, Talwalker S, Schwartz BD, Isakson PC, Geis GS. 1998. Preliminary study of the safety and efficacy of SC 58635, a novel cyclooxygenase 2 inhibitor: Efficacy and safety in two placebo controlled trials in osteoarthritis and rheumatoid arthritis, and studies of gastrointestinal and platelet effects. Arthritis Rheum 41(9):1591–1602. 21. Kakumanu VK, Bansal AK. 2002. Enthalpy relaxation studies of celecoxib amorphous mixtures. Pharm Res 19(12):1873–1878. 22. Callahan JC, Cleary GW, Elefant M, Kaplan G, Kensler T, Nash RA. 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev Ind Pharm 8(3):355–369. 23. Testa B, Carrupt PA, Gaillard P, Billois F, Weber P. 1996. Lipophilicity in molecular modeling. Pharm Res 13(3):335–343. 24. Modi SR, Dantuluri AK, Puri V, Pawar YB, Nandekar P, Sangamwar AT, Perumalla SR, Sun CC, Bansal AK. 2013. Impact of crystal habit on biopharmaceutical performance of celecoxib. Cryst Growth Des 13(7):2824–2832. 25. Puri V, Dantuluri AK, Kumar M, Karar N, Bansal AK. 2010. Wettability and surface chemistry of crystalline and amorphous forms of a poorly water soluble drug. Eur J Pharm Sci 40(2):84–93. 26. Fedors RF. 1974. A method for estimating both the solubility parameters and molar volumes of liquids. Polym Eng Sci 14(2):147–154. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
27. Hancock BC, York P, Rowe RC. 1997. The use of solubility parameters in pharmaceutical dosage form design. Int J Pharm 148(1):1–21. 28. Boyer RF. 1977. Encyclopedia of polymer science and technology; Mark HF, Bikales NM, Eds. New York: Interscience Publishers, pp 745. 29. Tong P, Zografi G. 1991. A study of amorphous molecular dispersions of indomethacin and its sodium salt. J Pharm Sci 90:1991–2004. 30. Shamblin S, Taylor LS, Zografi G. 1998. Mixing behaviour of colyophilized binary systems. J Pharm Sci 87:694–701.
DOI 10.1002/jps.23999
9
31. Hallbrucker A, Mayer E, Johari GP. 1989. The heat capacity and glass transition of hyperquenched glassy water. Philos Mag B 60(2):179–187. 32. Kaushal AM, Bansal AK. 2008. Thermodynamic behavior of glassy state of structurally related compounds. E J Pharm Biopharm 69:1067– 1076. 33. Zurek WH. 1989. Algorithmic randomness and physical entropy. Phys Rev A 40(8):4731.
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
Effect of Different “States” of Sorbed Water on Amorphous Celecoxib GANESH SHETE,1 SWATHI KUNCHAM,2 VIBHA PURI,3 RAHUL P. GANGWAL,4 ABHAY T. SANGAMWAR,4 ARVIND KUMAR BANSAL1 1
Department of Pharmaceutics, National Institute of Pharmaceutical Education and Research (NIPER), S.A.S. Nagar, Punjab 160062, India Generics FR&D, Dr. Reddy’s Laboratories, Bachupally, Hyderabad 500072, India 3 Novartis-MIT Centre for continuous manufacturing, Cambridge, Massachusetts 02139 4 Department of Pharmacoinformatics, National Institute of Pharmaceutical Education and Research (NIPER), S.A.S. Nagar, Punjab 160062, India 2
Received 13 November 2013; revised 7 April 2014; accepted 14 April 2014 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23999 ABSTRACT: Glass transition temperature (Tg ) of an amorphous drug is a vital physical phenomenon that influences its visco-elastic properties, physical, and chemical stability. Water acts as a plasticizer for amorphous drugs thus increasing their recrystallization kinetics. This reduces the solubility advantage of an amorphous drug. Hence, there is an interest in understanding the relationship between water content and Tg of amorphous drug. We have studied the effect of “state” of sorbed water on Tg of amorphous celecoxib (ACLB). ACLB was allowed to sorb water at relative humidity of 33%, 53%, 75%, and 93%. ALCB showed biphasic sorption of water designated as “bound” and “solvent-like” state of water associated with ACLB. Molecular modeling studies provided deeper insights into the interaction of water with ACLB. A distinct co-relationship between the state of water and its plasticization capacity was observed. Bound state of water had a very profound effect on the fall in experimentally observed Tg (Tg-exp ) value. Solvent-like state of water had little impact on Tg-exp value. Tg of ACLB–water mixture was predicted by Gordon–Taylor equation (Tg-pre ). The deviations in Tg-exp and Tg-pre were correlated to volume C 2014 non-additivity and non-ideal mixing. This study has implications on the development of formulations based on amorphous forms. Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci Keywords: Amorphous; glass transition; water sorption; water in solids; molecular modeling
INTRODUCTION Amorphous solids offer several advantages in pharmaceuticals; however, their industrial application has remained limited because of their inherent tendency to convert to the stable crystalline state.1 Amorphous materials have higher solubility than their crystalline counterparts because of the disordered state and higher free energy. Effective “stabilization” of amorphous state can enable exploitation of their solubility advantage.2 Interaction of amorphous pharmaceuticals with water has always evoked interest. Water is ubiquitously present in pharmaceutical systems and more so in amorphous state because of their higher hygroscopicity. Water is a strong plasticizer with a glass transition temperature (Tg ) of −138◦ C.3 It has a deleterious effect on the amorphous substances as it lowers their Tg and accelerates the recrystallization process. Latter reduces the solubility advantage achievable from the amorphous systems.4 The impact of water on various properties of amorphous systems like Tg and recrystallization behavior has been studied on pharmaceuticals like sequinavir, lamotrigine mesylate, nifedipine, felodipine, sucrose, indomethacin, and lactose3,5–16 using numerous methods. However, reports investigating the effect of “state” of sorbed water on Tg are lacking in literature.
Correspondence to: Arvind Kumar 2214682ext2126; Fax: +91-172-2214692; [email protected])
Bansal (Telephone: +91-172E-mail: [email protected];
Journal of Pharmaceutical Sciences C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association
A review paper published in 1988 classified water sorption by amorphous solids into two different types: “bound state” and “solvent-like” state of water.17 “Bound water” involves specific hydration directly to the molecules of amorphous solid and affects its physical and chemical phenomena. At higher water contents, clusters are formed and a “solvent-like” state of water is generated. Because this water is not available directly in the vicinity of amorphous solid, it exhibits limited effects on the latter. The concepts proposed in the aforementioned reference had not been justified by experimental findings/examples.17 In contrast to the two-state model, Hancock et al.18 had suggested a “mutual miscibility” or “plasticizer” model to explain the water vapor sorption by amorphous material. This model envisages a continuum of chemical, physical, and energetic states.18 The study was based on hydrophilic polymers like polyvinylpyrrolidone (PVP) that allows significant interaction with water molecules. The Tg for drug–water binary system can be predicted using theoretical approaches like Fox equation or Gordon–Taylor equation. However, deviations of experimentally observed Tg (Tg-exp ) from theoretically predicted Tg (Tg-pre ) values are common. These deviations have been attributed to various reasons like molecular mobility,14 differences in molecular weight,19 free volume,5 and/or molecular size.3 The purpose of the present work is to investigate the effect of “state” of sorbed water on Tg-exp of amorphous celecoxib (ACLB). To the best of our knowledge, this is the first report on the impact of “state” of sorbed water on Tg-exp of an amorphous form of a low-molecular-weight pharmaceutical molecule. Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
1
2
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Powder X-Ray Diffraction Powder X-ray diffraction (PXRD) patterns of samples were recorded at room temperature on Bruker’s D8 Advance diffractometer (Karlsruhe, West Germany) with Cu K" radiation ˚ at 40 kV and 40 mA passing through nickel filter (1.54 A), with divergence slit (0.5◦ ), antiscattering slit (0.5◦ ), and receiving slit (0.1 mm). PXRD scans were performed with a step size of 0.01◦ and step time of 1 s over an angular range of 3◦ –40◦ 2θ. Obtained diffractograms were analyzed with DIFFRACplus EVA (version 9.0) diffraction software. Microscopy Figure 1. Chemical structure of celecoxib.
Further, comprehensive reasons for variability in Tg-exp and Tg-pre have been discussed, considering contributors like volume non-additivity and non-ideality in the mixing of water and ACLB. This understanding has significant implications on drug development using amorphous system.
MATERIALS AND METHODS Material The crystalline form of celecoxib (CLB) was purchased from Unichem Laboratories Ltd. (Raigad, India). The purity of the sample was confirmed to be greater than 99.0% by the certificate of analysis provided by Unichem Laboratories. CLB is a Biopharmaceutics Classification System class II drug.20 ACLB has been reported to have a Tg-exp of 51◦ C–57 ◦ C and mean molecular relaxation time of 2.79 h, as calculated by Kohlrausch–Williams–Watts equation (β = 0.274).21 Chemical structure of CLB is shown in Figure 1. Methods Preparation of ACLB In situ ACLB was generated using differential scanning calorimetry (DSC). Accurately weighed amount of CLB (about 5 mg) was taken in an aluminum pan and heated to 185◦ C at heating rate of 20◦ C/min, held isothermally for 1 min to ensure complete melting followed by quench cooling to 25◦ C at cooling rate of 20◦ C/min under dry nitrogen purge of 50 mL/min. HPLC analysis indicated that no chemical degradation occurred during the generation of ACLB. Ex situ ACLB was prepared in comparatively larger quantity to carry out water sorption studies. CLB was melted in R ; Associated a stainless steel beaker on a hot plate (Sonar Scientific Technologies, New Delhi, India), and subsequently quench cooled over crushed ice. The quench-cooled product was passed through British Standard Sieve 44 resulting in a particle size range of about 5–50 :m with a D90 of 40 :m. No devitrification occurred during this processing. HPLC analysis indicated that no chemical degradation occurred during the generation of ACLB. All operations required for ex situ ACLB generation were performed in a controlled humidity condition [< 15% relative humidity (RH)] and at ambient temperature. Initial water content of ACLB samples was found to lesser than 0.05% (w/w), as determined by Karl Fisher titrimetry (716 DMS Titrino; Metrohm Limited, Herisau, Switzerland). Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
Samples were visually observed under Leica DMLP polarized microscope (Leica Microsystems Wetzlar GmbH, Wetzlar, Germany) in a bright light and cross-polarized light mode. Photomicrographs were acquired using Leica DC 300 camera and analyzed using Linksys 32 software. Water Vapor Sorption For the assessment of the effect of sorbed water on Tg-exp , ACLB samples were exposed to 33%, 53%, 75%, and 93% RH in an open DSC pan at 25◦ C in a vacuum desiccator up to 5 h (h). The samples were withdrawn at different time points up to 5 h. Water sorption of ACLB was measured gravimetrically from the difference between the weight of the dry sample immediately after its preparation and that after humidification. All the weights were measured using Sartorius GC 2502 weighing balance (New York, USA). These samples were subsequently analyzed in DSC after hermetically sealing the DSC pans. All the measurements were carried out in triplicates. Differential Scanning Calorimetry Conventional DSC experiments were conducted using DSC Q2000 (TA Instruments, Delaware, USA) equipped with a refrigerated cooling system and operating with Universal Analysis 2000 software version 4.5A. The instrument was calibrated for heat flow and temperature with high purity standards of Indium and Zinc. An important concern in the study of the effect of water on amorphous drug is the determination of Tg-exp in DSC. Several reports are available in the literature regarding the use of DSCbased protocols like (1) a single heating run of 0◦ C–200◦ C and (2) two heating runs, wherein first heating run is carried out up to a temperature limit of Tg + 20◦ C to eliminate the thermal history of sample and second heating run is used to determine Tg-exp .11 Tg-exp values determined from either of the protocols did not show any significant differences in our study. Hence, a single heating run protocol was adopted. Samples were scanned at a heating rate of 20◦ C/min in hermetically sealed aluminum pans from 0◦ C to 200◦ C with nitrogen purging (50 mL/min). Tg-exp , Tc , and Tm have been reported as onset temperature values determined in first heating run. Water affects the Tg-exp of material only when it is in physical contact with the amorphous matrix. For this purpose, hermetically sealed DSC pans were used for the analysis. About 7– 10 mg sample was added into DSC pan to keep headspace minimum during sealing of pan and lid to avoid evaporation of water into headspace during heating. All the studies have been performed in triplicates and average values are reported. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
3
Figure 2. PXRD patterns of (i) CLB and (ii) ACLB generated by in situ and ex situ method. Inset depicts zoomed versions of PXRD patterns of in situ and ex situ ACLB.
Molecular Modeling All the molecular modeling calculations were performed using SYBYL 7.1 package installed on a Silicon Graphics Fuel Workstation running IRIX 6.5. The CLB starting ensemble was built from available crystal data. Random arrangement of 55 CLB molecules, simulating amorphous form, was modeled by molecular silverware method. The molecules were later minimized by applying molecular mechanics force field (MMFF94) with the conjugate gradient method. The minimization was terminated when the energy gradient convergence criterion of 0.05 kcal/mol was reached or when the 100 step minimization cycle was exceeded. The optimized model was further subjected to molecular dynamics (MD) simulation using simulated annealing approach. The system was heated to 185◦ C to transform the crystal into the liquid state, and equilibrated for 1000 femtoseconds (fs) at 185◦ C, and finally cooled to 25◦ C. The technique was analogous to experimental procedure adopted for preparing ACLB by melting the drug and suddenly cooling it. Molecular dynamics simulation was performed for simulation of water vapor sorption by ACLB. The lattice of 10 × ˚ 3 was created, with 0.0279 g/cm3 density of water 10 × 10 A vapor at 93% RH and at 25◦ C, by molecular silverware method. The constructed system was minimized by applying molecular mechanics force field (MMFF94) with the 100 minimization cycle of steepest decent method and 100 minimization cycle of DOI 10.1002/jps.23999
conjugate gradient method. Periodic boundary conditions were imposed in order to eliminate the surface effects. The systems were equilibrated using constant number of atoms, volume, and energy ensemble for 50 picoseconds (ps) with a 0.5 fs time step. Then, MD simulation was performed on minimized equilibrated water–ACLB system at 25◦ C under constant moles, volume, and temperature (NVT) condition. The number of steps of NVT MD simulation was 500 ps with time step of 1 fs and output frequency was after every 100 steps.
RESULTS Solid-State Characterization Figure 2 shows PXRD scans of CLB and ACLB generated by in situ and ex situ method. PXRD analysis of CLB revealed its crystalline nature. CLB exhibited peaks at 2θ values of 5.36◦ , 10.72◦ , 13.01◦ , 14.80◦ , 16.10◦ , 19.65◦ , 21.52◦ , 23.45◦ , 25.38◦ , 27.00◦ , and 29.59◦ corresponding to crystalline form III. ACLB generated by in situ and ex situ method exhibited halo pattern in PXRD at room temperature. DSC scans of CLB, in situ ACLB and ex situ ACLB are shown in Figure 3. CLB showed a single sharp endotherm corresponding to melting (Tm ) at 162.3◦ C. In situ generated ACLB exhibited a Tg-exp of 57.6◦ C. No recrystallization or melting event was Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
4
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 3. DSC traces of (i) CLB and (ii) ACLB generated by in situ and ex situ method.
evident in this sample. However, it was not possible to generate in situ sample in quantities sufficient for water sorption studies. Hence, ACLB was generated using an ex situ protocol. Ex situ generated ACLB exhibited a halo pattern in PXRD (Fig. 2) and absence of birefringence in polarized light microscopy studies (data not shown), thus establishing its amorphous nature. In DSC, ex situ ACLB exhibited a Tg-exp at 53.8◦ C and heat induced recrystallization (Tc ) at 126.1◦ C–152.6◦ C followed by a broad melting endotherm at 161.4◦ C. The recrystallization can be attributed to nonisothermal run in DSC. The ex situ ACLB was used for subsequent water sorption studies.
Water Sorption Behavior of ACLB Exposure of ACLB to 33%, 53%, 75%, and 93% RH conditions at 25◦ C in desiccators led to recrystallization of ACLB and the recrystallization rate increased from 33% to 93% RH condition, as was evident by birefringence observed under cross polarized light in microscope (data not shown). The minimum time required for onset of recrystallization at 93% RH (at which the recrystallization rate was maximum) was approximately 5 h. The presence of crystallinity in amorphous material is likely to change the response of amorphous phase toward water in terms of plasticization and recrystallization. To avoid these possibilities, all the water sorption studies at 33%, 53%, 75%, and 93% RH were performed only up to 5 h wherein the sample was completely in amorphous state. ACLB sorbed water with increasing rates at 33%, 53%, 75%, and 93% RH linearly. ACLB sorbed maximum water of 1.88% (w/w) at 93% RH and at 5 h. ACLB was found to be nonhygroscopic as per Callahan hygroscopicity classification.22 Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
Effect of Sorbed Water on Tg Nonisothermal heating run of ACLB with sorbed water (ACLB– water mixture) in DSC demonstrated phenomena of Tg-exp , Tc , and Tm . Table 1 lists Tg-exp , Tc , and Tm values for ACLB–water mixture. The theoretically expected value of Tg (Tg-pre ) was calculated by using Gordon–Taylor equation, which is given as follows3 : Tg−pre =
w1 Tg1 + kw2 Tg2 w1 + kw2
(1)
where Tg1 and Tg2 are Tg values of water and ACLB, respectively. w represents the weight fraction and k is ratio of free volumes given by, k=
1 Tg1 2 Tg2
(2)
where ρ 1 and ρ 2 indicate the true densities of water and ACLB, respectively. Figure 4 shows the behavior of Tg-exp and Tg-pre with respect to sorbed water. The Tg-exp showed a fall with increasing water content because of the plasticizing effect of water. Interestingly, Tg-exp demonstrated steepest fall till water content of 1% (w/w). Thereafter, a little effect of water on Tg-exp was observed. Reduction in Tc values from 126◦ C to 76◦ C was observed with increasing water content. DSC analysis was performed on samples that had not shown recrystallization during storage and thus residual crystallinity did not contribute to Tc . Thus, the observed reduction in Tc was contributed only by the water in the amorphous matrix.19 DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Table 1.
Tg-exp , Tg-pre , Tc , and Tm Values for ACLB–Water Mixture
Water Content (%, w/w) Tg-exp (◦ C)a Tg-pre (◦ C)b Tc (◦ C)a Tm (◦ C)a 0 0.17 0.20 0.23 0.24 0.33 0.39 0.42 0.46 0.54 0.60 0.66 0.68 0.74 0.79 0.85 0.90 0.95 1.00 1.13 1.27 1.40 1.88 a b
53.8 52.4 51.3 50.6 47.5 46.8 45.1 44.5 43.2 42.5 41.9 41.8 41.7 40.8 39.1 37.1 36.2 35.1 34.9 34.5 34.2 34.1 31.0
53.8 53.8 51.3 51.2 51.0 50.9 50.4 50.1 49.9 49.6 49.2 48.8 48.5 48.4 48.0 47.7 47.4 47.2 46.9 46.6 45.9 45.1 44.3
126.1 108.2 105.8 103.0 103.9 102.0 101.8 101.4 99.8 98.5 95.4 90.4 89.6 88.0 87.5 85.6 84.3 83.8 82.8 76.9 76.5 76.8 76.9
161.4 160.5 160.3 160.6 160.2 161.7 160.4 160.9 160.2 160.1 160.3 161.8 160.4 160.5 161.4 160.3 160.1 160.3 161.8 160.4 160.5 160.5 160.7
Mean of experimentally observed values for n=3. Predicted values by Gordon–Taylor equation.
DISCUSSION Mechanism of Water Sorption by ACLB Matrix The sorption of water up to 1.00% (w/w) resulted in drop of Tg-exp values, thus pointing toward close interaction of water with ACLB. However, beyond 1% (w/w) water content, no appreciable change in Tg-exp was observed. Our findings do not support concept of “continuum model” of water sorption proposed by
5
Hancock et al.18 The observations and proposal for continuum model was based on the experiments carried out with PVP, which is a hydrophilic polymer. ACLB possesses relatively hydrophobic molecular environment (discussed in detail in the preceding section) and exhibited a behavior that closely resembled with concept proposed by Prof. Zografi.17 Hence, in the present work, the sorption of water up to 1.00% (w/w) can be designated as “bound” state of water. The “bound” water is in closest vicinity of solids and directly hydrates the molecules of solid matrix. This type of water affects the molecular level properties of solids as well as chemical reactions. The second phase of water sorption by ACLB from 1% to 1.88% (w/w) can be designated as the “solvent-like” state of water, which was found to have negligible effect on Tg-exp . At higher water content, water molecules form clusters and give rise to “solventlike” state.17 These types of clusters cannot effectively contact the hydrophobic amorphous matrix and thus exert negligible effects on amorphous material. The molecular environment possessed by the pharmaceutical material determines its interaction with water. In case of a low-molecular-weight hydrophobic organic pharmaceutical compound like ACLB, water can interact only up to a limited extent, as governed by the hydrophilic sites in the molecular structure. A limited interaction is possible between molecules of water and ACLB, after which continuum is disturbed and clusters of water molecules are formed over amorphous bed.18 Further, CLB is a hydrophobic compound and it is very less likely that it will show capillary condensation. The “mutual miscibility” or “continuum” model suggested by Hancock et al.18 thus seems to be inapplicable in the case of ACLB–water system. Molecular Understanding of Interaction of Water with ACLB In the last few decades, molecular modeling has been utilized in rationalizing the experimental observations and provides deeper insights into “difficult-to-explain” events. We carried
Figure 4. Comparison of fall in Tg-exp () and Tg-pre (О) with increasing water content for ACLB–water mixture. DOI 10.1002/jps.23999
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
6
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 5. (a) Computer simulation of molecular arrangement of ACLB, (b) interaction of water molecules with ACLB bed, (c) pictorial depiction of states of water associated with ACLB matrix, and (d) hydrogen bonding interactions between water and CLB molecules; based on the special constraints, all CLB molecules do not show all types of interactions.
out the MD study to understand the moisture sorption behavior by ACLB. Figure 5a shows ACLB matrix simulated on the computer and Figure 5b shows interactions of ACLB with water. Water was preferentially found to interact with hydrophilic sites of ACLB through H-bonding. This type of sorbed water can be designated as “bound” type of water. After all the hydrophilic sites are saturated, the water molecules started to form clusters around already sorbed water that can be termed as “solvent-like” state of water. Figure 5c depicts bound and solvent-like states of water associated with ACLB matrix. Molecular lipophilic surface potential (MLSP) is a measure of lipophilicity of a molecule, which describes combined lipophilic influence of all fragments contained in it.23 Evaluation of MLSP of crystalline CLB revealed contribution of –SO2 NH2 group toward hydrophilicity and contribution of lipophilic methyl phenyl moiety toward lipophilicity.24 Findings of molecular modeling studies provided interesting insights. Hydrogen bonding (H-bonding) was observed between –SO2 NH2 group of CLB and water molecules. Some of CLB molecules also showed additional H-bond interaction with water through –NH group of Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
pyrazole ring. This H-bonding between ACLB and bound water affects the solid-state properties of the former like Tg and recrystallization tendency. The H-bonding interactions between water and CLB molecules are shown in Figure 5d. Evidence of H-bonds between ACLB and water has also been established earlier from differential heat of sorption at 25◦ C/54% RH (155.04 ± 5.02 kJ/mol), which was found to be greater than the heat of condensation of water (44 kJ/mol).25 We were able to unveil the average interaction occurring between single CLB and water molecules (Fig. 5d). However, not all CLB molecules in amorphous phase participated in similar interactions with water molecules. Some of the domains of CLB molecules involved in interactions with water are “preengaged” because of their interaction with neighboring CLB molecules. Amorphous phase lacks long-range order, but possesses short-range order. This short-range order presents differential molecular environment around various CLB molecules. Although from a molecular perspective, each CLB molecule offers possibility of interacting with five water molecules, but the supramolecular environment of amorphous phase forces different CLB molecules to interact with water in a different manner. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
This prevented us from quantifying the average number of water molecules that interact with single CLB molecules. Experimental evidence for limited interaction of water with ACLB has been reported in our previous publications.25 A significant difference between solubility parameter (δ t ) values of CLB (21.01) and water (47.9) suggested their immiscibility and thus limited molecular interaction. δ t of CLB was calculated by the group contribution method and found to be 21.01,26,27 whereas δ t of water was taken from literature as 47.9.28 Mixing of Water with ACLB The plasticizing effect of water on amorphous phase of drugs can be predicted by calculating Tg-pre of a drug–water mixture using Gordon–Taylor equation. Tg-exp values were found be lower than that of Tg-pre . The deviation between Tg-exp and Tg-pre became prominent with the increase in the water content up to 1.0% (w/w). Thereafter, the deviation narrowed down at sorbed values above 1% (w/w), as shown in Figure 4. The differences observed between Tg-exp and Tg-pre can be explained by the thermodynamic calculations and discussion provided below. The thermodynamic calculations were meant to further explore the mixing behavior of water with ACLB. Gordon–Taylor equation is based on the free volume theory and assumes ideal mixing. It describes the behavior of a “mis-
7
cible” amorphous mixture of a binary system. However, most of the amorphous molecular dispersions are supersaturated “solutions,” as during formation, materials are trapped in a glass, with no measurable phase separation of individual phases. In DSC measurements, miscibility is evidenced by a single Tg .29 The constant k of the equation can be calculated from known parameters and allows use of weight fraction to describe compositions in such systems. Ideal behavior of mixtures can be predicted in terms of the enthalpy change of mixing which is zero for ideal mixing, and negative or positive for non-ideal mixing. The deviation between Tg-exp and Tg-pre indicated nonadditivity of volume because of the non-uniform distribution of the free volume between ACLB and water, and huge difference in their Tg values. These deviations are a result of molecular interactions between ACLB and water. As evidenced by molecular modeling studies and reports from literature,25 ACLB and water can form H-bonds. However, a weaker set of inter-molecular interactions between CLB and water as compared with CLB– CLB + water–water leads to volume expansion and volume non-additivity.30 The phenomenon of volume non-additivity is usually associated with the non-ideality of mixing. Non-ideality of mixing can be predicted by determination of excess enthalpy (HE ), entropy (SE ), and free energy (GE ) of mixing.30 These
Figure 6. Enthalpy, entropy, and free energy of mixing for ACLB–water mixture. DOI 10.1002/jps.23999
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
8
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
parameters were calculated with respect to % (w/w) water content and are shown in Figure 6. Average Tg-exp values were used for thermodynamic calculations. Cp values at Tg for water and ACLB were used from literature as 0.088931 and 0.4421 J/K32 , respectively. The net enthalpy of mixing increased with the increase in water content and indicated disruption of H-bonds, thus demanding more energy for interaction. Simultaneously, the breakage of “native order” of disordered amorphous matrix of ACLB by the addition of water increased the net entropy of mixing.33 Further, free energy represents the difference in the free energy of the mixture at Tg-exp and free energy of the mixture at temperature T assuming that the entropy of mixing is purely combinatorial and the mixing is non-thermal. Free energy decreased because of the relatively higher increase in the factor of Tg-exp as compared with the HE . The evaluation of enthalpy, entropy, and free energy of mixing indicated non-ideal mixing behavior of ACLB–water system. Biphasic curves were observed with leveling off in the curves after 1% (w/w) sorbed water content, indicating that the nonideality of mixing had attained its peak value at 1% (w/w) sorbed water. This further supports the fact that water is in direct contact with ACLB matrix only up to 1% (w/w) and represents “bound” state of water. The leveling of the curves after 1% (w/w) indicates that the non-interactive nature of water represents its “solvent-like” state.
CONCLUSIONS The interaction of water with pharmaceutical solids is critical during processing and shelf life stability. Water is ubiquitously present in drugs and excipients and can impact solid phase behavior of ingredients. Although total water content is important, the state in which water is present is even more critical. Water present in the “bound” state participates in molecular interactions most profoundly thus affecting the solid state behavior like Tg and recrystallization tendency. Critical assessment of the sorbate properties provides insights into its interaction with water. Volume non-additivity and non-ideal mixing provide the underlying thermodynamic basis for these interactions. Additional studies on materials with diverse physicochemical properties can generate guidelines for the role of water in performance of drug products, containing amorphous pharmaceuticals.
ACKNOWLEDGMENTS The authors thank Dr. Bruno Hancock and Prof. George Zografi for their valuable inputs and helpful discussions. Ganesh Shete and Rahul P. Gangwal acknowledge the financial support from Council of Scientific and Industrial Research, Government of India.
REFERENCES 1. Craig DQ, Royall PG, Kett VL, Hopton ML. 1999. The relevance of the amorphous state to pharmaceutical dosage forms: Glassy drugs and freeze dried systems. Int J Pharm 179(2):179–207. 2. Hancock BC, Parks M. 2000. What is the true solubility advantage for amorphous pharmaceuticals? Pharm Res 17:397–403. 3. Hancock BC, Zografi G. 1994. The relationship between the glass transition temperature and the water content of amorphous pharmaceutical solids. Pharm Res 11(4):471–477. Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
4. Buckton G, Patricia D. 1999. Assessment of disorder in crystalline powders-a review of analytical techniques and their application. Int J Pharm 179:141–158. 5. Andronis V, Yoshioka M, Zografi G. 1997. Effects of sorbed water on the crystallization of indomethacin from the amorphous state. J Pharm Sci 86:346–351. 6. Buckton G, Darcy P. 1996. Water mobility in amorphous lactose below and close to the glass transition temperature. Int J Pharm 136:141– 146. 7. Burnett DJ, Thielmann F, Sokoloski T, Brum J. 2006. Investigating the moisture-induced crystallization kinetics of spray-dried lactose. Int J Pharm 313(1–2):23–28. 8. Grigera JR, Viviana B. 2008. Molecular mobility and ageing of sugar glasses. Food Chem 106:1314–1317. 9. Hu X, Kaplan D, Cebe P. 2007. Effect of water on the thermal properties of silk fibroin. Thermochim Acta 461:137–144. 10. Imamura K, Atsushi F, Sakaura K, Sugita T, Sakiyama T. 2002. Water sorption and glass transition behaviors of freeze-dried sucrosedextran mixtures. J Pharm Sci 91:2175–2181. 11. Konno H, Taylor LS. 2008. Ability of different polymers to inhibit the crystallization of amorphous felodipine in the presence of moisture. Pharm Res 25(4):969–978. 12. Pittia P, Sacchetti G. 2008. Antiplasticization effect of water in amorphous foods. A review. Food Chem 106:1417–1427. 13. Royall PG, Craig DQ, Doherty C. 1999. Characterisation of moisture uptake effects on the glass transitional behaviour of an amorphous drug using modulated temperature DSC. Int J Pharm 192(1):39–46. 14. Schmitt E, Davis CW, Long ST. 1996. Moisture-dependent crystallization of amorphous lamotrigine mesylate. J Pharm Sci 85(11):1215– 1219. 15. Schubnell M, Schawe JE. 2001. Quantitative determination of the specific heat and the glass transition of moist samples by temperature modulated differential scanning calorimetry. Int J Pharm 217(1– 2):173–181. 16. Shamblin SL, Zografi G. 1999. The effects of absorbed water on the properties of amorphous mixtures containing sucrose. Pharm Res 16:1119–1124. 17. Zografi G. 1988. States of water associated with solids. Drug Dev Ind Pharm 14:1905–1926. 18. Hancock BC, Zografi G. 1993. The use of solution theories for predicting water vapor absorption by amorphous pharmaceutical solids: A test of the Flory–Huggins and Vrentas models. Pharm Res 10(9):1262– 1267. 19. Saleki-Gerhardt A, Zografi G. 1994. Non-isothermal and isothermal crystallization of sucrose from the amorphous state. Pharm Res 11:1166–1173. 20. Simon LS, Lanza FL, Lipsky PE, Hubbard RC, Talwalker S, Schwartz BD, Isakson PC, Geis GS. 1998. Preliminary study of the safety and efficacy of SC 58635, a novel cyclooxygenase 2 inhibitor: Efficacy and safety in two placebo controlled trials in osteoarthritis and rheumatoid arthritis, and studies of gastrointestinal and platelet effects. Arthritis Rheum 41(9):1591–1602. 21. Kakumanu VK, Bansal AK. 2002. Enthalpy relaxation studies of celecoxib amorphous mixtures. Pharm Res 19(12):1873–1878. 22. Callahan JC, Cleary GW, Elefant M, Kaplan G, Kensler T, Nash RA. 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev Ind Pharm 8(3):355–369. 23. Testa B, Carrupt PA, Gaillard P, Billois F, Weber P. 1996. Lipophilicity in molecular modeling. Pharm Res 13(3):335–343. 24. Modi SR, Dantuluri AK, Puri V, Pawar YB, Nandekar P, Sangamwar AT, Perumalla SR, Sun CC, Bansal AK. 2013. Impact of crystal habit on biopharmaceutical performance of celecoxib. Cryst Growth Des 13(7):2824–2832. 25. Puri V, Dantuluri AK, Kumar M, Karar N, Bansal AK. 2010. Wettability and surface chemistry of crystalline and amorphous forms of a poorly water soluble drug. Eur J Pharm Sci 40(2):84–93. 26. Fedors RF. 1974. A method for estimating both the solubility parameters and molar volumes of liquids. Polym Eng Sci 14(2):147–154. DOI 10.1002/jps.23999
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
27. Hancock BC, York P, Rowe RC. 1997. The use of solubility parameters in pharmaceutical dosage form design. Int J Pharm 148(1):1–21. 28. Boyer RF. 1977. Encyclopedia of polymer science and technology; Mark HF, Bikales NM, Eds. New York: Interscience Publishers, pp 745. 29. Tong P, Zografi G. 1991. A study of amorphous molecular dispersions of indomethacin and its sodium salt. J Pharm Sci 90:1991–2004. 30. Shamblin S, Taylor LS, Zografi G. 1998. Mixing behaviour of colyophilized binary systems. J Pharm Sci 87:694–701.
DOI 10.1002/jps.23999
9
31. Hallbrucker A, Mayer E, Johari GP. 1989. The heat capacity and glass transition of hyperquenched glassy water. Philos Mag B 60(2):179–187. 32. Kaushal AM, Bansal AK. 2008. Thermodynamic behavior of glassy state of structurally related compounds. E J Pharm Biopharm 69:1067– 1076. 33. Zurek WH. 1989. Algorithmic randomness and physical entropy. Phys Rev A 40(8):4731.
Shete et al., JOURNAL OF PHARMACEUTICAL SCIENCES
