A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin

G Model IJP-12600; No. of Pages 8

ARTICLE IN PRESS International Journal of Pharmaceutics xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm

A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin Korbinian Löbmann a , Riikka Laitinen a , Holger Grohganz b , Clare Strachan a , Thomas Rades a,b,∗ , Keith C. Gordon c a

School of Pharmacy, University of Otago, Dunedin, New Zealand Department of Pharmacy, University of Copenhagen, Copenhagen, Denmark c Department of Chemistry and MacDiarmid Institute for Advanced Materials and Nanotechnology, University of Otago, Dunedin, New Zealand b

a r t i c l e

i n f o

Article history: Received 6 March 2012 Received in revised form 7 May 2012 Accepted 11 May 2012 Available online xxx Keywords: Density functional theory Quantum mechanical chemistry Co-amorphous Naproxen-indomethacin Infrared spectroscopy

a b s t r a c t Co-amorphous drug systems were recently introduced as potential drug delivery systems for poorly water soluble drugs in order to overcome problems associated with amorphous materials. The improved physical stability and dissolution of these systems was attributed to molecular interactions between the co-amorphous partners, such as hydrogen bonds. However, molecular level characterization with vibrational spectroscopy of even the amorphous drugs alone presents a significant challenge. This becomes even more complicated when more than one compound is present in the material under investigation. In this study, the co-amorphous drug mixture containing naproxen (NAP) and indomethacin (IND) was investigated using infrared spectroscopy (IR) and quantum mechanical calculations. The structures of both drugs were optimized as monomer, homodimer and heterodimer using density functional theory and used for the calculation of IR spectra. Conformational analysis confirmed that the optimized structures were suitable for the theoretical prediction of the spectra. Vibrational modes from the calculation could be matched with experimentally observed spectra for crystalline and amorphous NAP and IND, and it could be shown that both drugs exist as homodimers in their respective individual amorphous form. With the results from the experimental single amorphous drugs and theoretical homodimers, a detailed analysis of the experimental co-amorphous and theoretical heterodimer spectra was performed and evaluated. It is suggested that NAP and IND exist as heterodimers in the co-amorphous mixture when quench cooled together from the melt in a 1:1 molar ratio. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Low aqueous solubility is a major concern for many drugs, and this problem is likely to increase in the future (Aaltonen and Rades, 2009; Müllertz et al., 2010). Poor water solubility is of great importance, as it may result in a low bioavailability of the drug especially when formulated as solid dosage forms (e.g. tablets, capsules) since their bioavailability relies mainly on the aqueous dissolution of the drug in the GI tract. Transforming a crystalline drug into its more soluble amorphous counterpart is one way to overcome this limitation (Aaltonen and Rades, 2009; Hancock and Zografi, 1997). In comparison to the well-defined three-dimensional order in the crystal lattice, the amorphous state is characterized by the absence of a long range molecular order. Nevertheless, molecules can still exhibit short range order, for

∗ Corresponding author at: School of Pharmacy, University of Otago, P.O. Box 913, Dunedin 9054, New Zealand. Tel.: +64 3 479 5410; fax: +64 3 479 7034. E-mail addresses: [email protected], [email protected] (T. Rades).

example by forming dimers through hydrogen bonding (Taylor and Zografi, 1997). Recently, co-amorphous systems have been found to be a promising way to stabilize the amorphous state (Allesø et al., 2009; Chieng et al., 2009; Löbmann et al., 2011, 2012). By the combination of two drugs to form a single phase amorphous system, the physical stability and aqueous dissolution could be improved over the individual amorphous drugs. Intermolecular interactions between the two drugs present in the system have been suggested to be responsible for this improvement. In our recent study on the coamorphous drug–drug combination between the two non-steroidal anti inflammatory drugs indomethacin (IND) and naproxen (NAP) it was shown that the co-amorphous 1:1 molar ratio exhibited a significant higher physical stability than the respective amorphous individual drugs and co-amorphous mixtures at molar ratios of 2:1 and 1:2 (Löbmann et al., 2011). Furthermore, intrinsic dissolution showed a synchronized release for the co-amorphous drugs at the 1:1 ratio and was significantly faster in comparison to the individual crystalline or amorphous (in case of IND) compounds. The dissolution rate of individual amorphous NAP could not be established

0378-5173/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8 2

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

due to the low physical stability of amorphous NAP. It was suggested that the changes in dissolution behavior might be explained by the formation of a heterodimer between both drugs, and that the heterodimer was responsible for the synchronized release and the increased physical stability. The synchronized release might also be responsible for the faster dissolution behavior of IND in the co-amorphous mixture. In order to obtain information on potential interactions in amorphous systems, vibrational spectroscopy provides insight into the molecular arrangement in amorphous systems. As spectroscopic methods probe the molecular level, changes in the molecular environment (e.g., H-bonding or ␲–␲ interactions) due to solid state changes may be detected as a result of altered vibrations of functional groups involved in these interactions (Chieng et al., 2011). However, these changes might only be minor or lost in the complexity of the spectra. Recently, quantum mechanical chemistry has become increasingly popular in the pharmaceutical field, especially due to its usefulness in the interpretation of vibrational spectra (Gordon et al., 2007). In addition, quantum mechanical calculations have been applied for example to determine the molecular structure with respect to the conformational minima of a drug molecule (Borba et al., 2009), however in most of the cases analysis has been simply applied for band assignment of the experimental infrared (IR) and Raman spectra of drugs (Ali et al., 2007; Borba et al., 2009; Hu et al., 2010; Iliescu et al., 2004; Jubert et al., 2006; Mishra et al., 2008; Sagdinc et al., 2007; Srivastava et al., 2010). Other studies have focused on spectra of various crystalline, polymorphic forms of drug molecules. Spectral differences in carbamazepine polymorphs I and III, for example, could be correlated with differences in the hydrogen bonding in the respective polymorphs (Strachan et al., 2004). Similar findings were obtained for chlorpropamide polymorphs. Again, differences were associated mainly to intermolecular interactions rather than to a different molecular conformation (Chesalov et al., 2008). The formation of dimers in the solid state has also been shown for other crystalline drugs, such as ibuprofen (Vueba et al., 2008), indomethacin (Strachan et al., 2007), ketoprofen (Vueba et al., 2006), and olanzapine (Ayala et al., 2006). Another study on budesonide revealed the epimeric contribution of the R and S epimer in the IR spectra (Ali et al., 2010). For theophylline, caffeine and theobromine, hydrogen bonding could be determined in anhydrate and hydrate forms with the help of computational chemistry (Nolasco et al., 2006). In addition to the computational and spectroscopic approach, XRPD was used in most of the studies to probe the crystal lattice as an additional tool. Thus, an insight on the molecular arrangement and potential molecular interactions within a crystal could be identified. Especially in the case of amorphous materials however, quantum mechanical calculations can become an invaluable tool to gain further understanding of the near range order of amorphous systems and to help interpret and support experimentally obtained spectra. Strachan et al. for example have shown that the dimer structure in crystalline ␥-indomethacin undergoes only little disruption upon transformation into its amorphous phase (Strachan et al., 2007). In comparison, for fenofibrate, which is a non hydrogen-bonded drug molecule with only weak intermolecular interactions (i.e. ␲–␲ interactions) in its crystalline state, Heinz et al. (2009a) suggested that these weak interactions can easily be disturbed, and the disruption of conformation and molecular orientation of fenofibrate is likely to be more random in its amorphous state. All the above mentioned computational studies dealt only with discrete drug molecules of one kind and their respective molecular near range order in various polymorphic forms or in the amorphous state. The situation becomes more complex in mixed systems, for example in co-amorphous drug combinations. Little is known about

Fig. 1. Molecular structures with numbering depicted for IND (A) and NAP (B). Subscript letters i and n are used to differentiate the atoms of both drugs.

the molecular arrangement of these systems and quantum mechanical calculations could provide a clearer insight. In this study, quantum mechanical chemistry was used to gain a better understanding on the near range order in co-amorphous IND and NAP. The main interest was to investigate the possible formation of a heterodimer. Firstly, a geometrical optimization of the drug molecules as single molecules, homodimers (IND–IND and NAP–NAP) and heterodimer (IND–NAP) was conducted. Secondly, calculations of the vibrational bands and band assignment were performed and compared the experimentally obtained FTIR spectra. To our knowledge, this is the first time that quantum mechanical calculations in combination with FTIR spectroscopy are used to interpret the molecular near range order of a co-amorphous drug combination. 2. Materials and methods 2.1. Materials (IND, ␥ polymorph; M = 357.79 g/mol; Indomethacin Tm = 162.0 ◦ C) and naproxen (NAP, M = 230.26 g/mol; Tm = 158.1 ◦ C) were sourced from Sigma–Aldrich, USA and Divis Laboratories, Ltd. USA, respectively. Molecular structures and numbering of the drug molecules are depicted in Fig. 1. The numbering system follows

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

3

that used by Kim and Song (1984) and Kistenmacher and Marsh (1972) for NAP and IND, respectively. Subscript letters i (for IND) and n (for NAP) are added in the numbering system in order to differentiate the individual atoms of both drugs.

2.2. Methods 2.2.1. Sample preparation Amorphous samples (IND, NAP and a 1:1 molar mixture of both drugs) were prepared by melting them in aluminum dishes (volume approx. 100 mL; bottom diameter: 8 cm) in a preheated oven at 168 ◦ C for 5 min and subsequent quench cooling by pouring liquid nitrogen onto the samples. Briefly, the sample dishes containing the molten compounds were placed in a desiccator containing phosphorus pentoxide, followed by pouring liquid nitrogen into the dishes. The desiccator was immediately closed afterwards in order to avoid moisture sorption onto the sample. The liquid nitrogen was allowed to evaporate before proceeding with the measurements of the infrared spectra. Before quench cooling of the 1:1 molar ratio, a physical mixture was prepared by gently mixing 1000 mg of the drugs (1:1 molar ratio; 608.1 mg IND and 391.6 mg NAP) with mortar and pestle for 60 s.

2.2.2. Attenuated total reflectance FT-infrared (ATR-FTIR) spectroscopy IR spectroscopy was performed using a Varian 3100 FTIR (Excalibur series) attached with an attenuated total reflectance accessory (GladiATR, Piketech, WI, USA). ATR-FTIR is a suitable set-up for structure analysis despite the fast recrystallization of NAP at room temperature, as the ATR crystal and surrounding plate could be cooled down with ice prior to the measurements to ensure experimental conditions below the glass transition temperature of NAP (Tg = 5.04 ◦ C) (Löbmann et al., 2011). Spectra of the crystalline drugs were obtained by using the powdered sample as received. Spectra were recorded over a range of 4000–400 cm−1 and a mean of 64 scans at a resolution of 4 cm−1 . Data analysis was performed on Varian Resolution Pro, v4.1.0.101 software and graphs were plotted using Origin (version 8.5.1, OriginLab Corporation, Northampton, MA, USA).

2.2.3. Computational studies Quantum mechanical modeling was performed on the individual drug monomers, homodimers and heterodimer (Fig. 2). As a starting point, the single drug molecules were determined using the bond parameters, angles and dihedral angles from the respective crystallographic data for NAP and ␥-IND. For the homodimer calculations, the single drug molecules were duplicated and arranged to form dimers. For the heterodimer a similar approach was performed by using one of the individual drug molecules each. Calculations were performed using the Gaussian 09 suite of programs (Frisch et al., 2009). Vibrational modes were determined using density functional theory (DFT) calculations (B3-LYP functional, 6-31G(d) basis set). The automated calculation by the software firstly optimized the conformations of the monomer and dimer structures. Secondly, the vibrational frequency calculations were determined to obtain the IR intensities. The predicted vibrational modes were scaled with the recommended factor of 0.9614 (Scott and Radom, 1996) using GaussSum 2.2 (O’Boyle et al., 2008) software and visualized using GaussView (v. 5.0, Gaussian, Inc., Wallingford, CT, USA).

Fig. 2. Molecular structures with hydrogen bonding between the carboxylic groups of the IND homodimer (A), NAP homodimer (B), and heterodimer of both drugs (C).

3. Results and discussion 3.1. Conformational analysis Kim et al. (1987) and Kistenmacher and Marsh (1972) have carried out crystallographic studies on NAP and ␥-IND, respectively. Their results were taken as a starting point for the optimization of the two drugs. No major differences in bond length and bond angles were found in the optimized monomer and dimer structures of both molecules compared to the respective experimental data obtained by Kim et al. (1987) and Kistenmacher and Marsh (1972) (data not shown). Therefore, these geometric parameters of crystalline NAP and IND were close to the lowest energy as calculated for the structures. This is not surprising since changes in bond length and angles are associated with large amounts of energy (Bernstein and Hagler, 1978), resulting in similar low energy conformation in both

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8 4

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

theoretical and experimental data. However, variations in dihedral angles between the calculated and experimental structures were observed. This can be explained by the low energy involved in torsional changes, i.e. single bond rotations (Bernstein and Hagler, 1978). For NAP, the main difference was found in the dihedral angles along the Cn 2 Cn 9 axis (Table 1), the connection between the isopropionic acid moiety and the naphthalene structure of the molecule. Here, the optimized structures were slightly rotated along this axis when compared to the crystal conformation. The molecular arrangement in the NAP crystal involves the incorporation in a rigid lattice where each molecule is surrounded by neighbor molecules and the molecules are arranged to build a chain through intermolecular hydrogen bonding (Kim and Song, 1984). In comparison, the calculated monomer was not involved in any hydrogen bonding and the dimer structures only involved one interaction, the cyclic hydrogen bonding. Further, in the rigid lattice of the crystal, single bond rotations can occur in order to fit the molecules into the crystal network (Bar and Bernstein, 1985). With other molecules similar observations with dihedral angles in quantum mechanical calculations have been reported (Heinz et al., 2009a; Strachan et al., 2004, 2007). In our previous study, it was suggested that NAP forms a homodimer once it is transferred into its amorphous form and a heterodimer with IND in the co-amorphous form (Löbmann et al., 2011). Therefore, in this study the focus was on the dimer formation. Unfortunately, it was not possible to optimize the NAP molecules in a chain-like structure similar to the crystal structure. The energetic minimum calculation for that conformation changed the molecular chain arrangement since the software does not take the crystal lattice into account. Therefore, prediction of an IR spectrum of naproxen arranged in a chain-like manner similar to its crystal structure could not be modeled. The torsional angles along the Cn 10 Cn 9 axis for the calculated NAP monomer and the experimental data were nearly identical (Table 1, line N5-8). Therefore, the molecules in the crystalline structure are close to the energetic minimum for this particular dihedral angle. Upon formation of the dimer structures these angles change by approximately 5◦ . Again, this can be explained by the differences in the crystal lattice compared to the dimer model in this study as mentioned above. For IND, similar observations on the dihedral angles were made. Again, the main difference was a single bond rotation, in particular along the Ci 18 Ci 19 axis (Table 2, line I5 & 6), i.e. the conformation of the carboxylic acid group. This group is involved in the cyclic hydrogen bonding of the dimer and differences can be explained by the same arguments as mentioned above. However, compared to NAP, the molecules do exist as dimers in the crystal structure of ␥IND (Kistenmacher and Marsh, 1972) and dimers are also thought to be formed in the amorphous form (Taylor and Zografi, 1997). Similar findings are also reported in a computational study on IND by Strachan et al. (2007). In order to further investigate the accuracy of the calculations, also the predicted intermolecular distances in the modeled hydrogen-bonded dimer structures were compared with the crystallographic data. The Oi 3(a) Oi 4(b) distance in the optimized IND ˚ and the On 2(a) On 1(b) distance in the calhomodimer (2.682 A) ˚ are close to the distances in culated NAP homodimer (2.693 A) ˚ and NAP (2.681 A), ˚ respectively. Intercrystalline IND (2.669 A) estingly, in the optimized heterodimer both values from the calculated homodimers appear with molecular distances of 2.682 A˚ and 2.693 A˚ for Oi 3(a) On 1(b) and On 2(b) Oi 4(a), respectively. Overall, the geometry of the optimized monomer and dimer structures of NAP and IND are similar to those in their respective crystalline forms. Small changes were found in the torsional angles of single bonds for the optimized molecules in the monomer,

Fig. 3. IR spectra of experimental crystalline NAP (a), experimental amorphous NAP (b), calculated NAP homodimer (c), and calculated NAP monomer (d) between 1000 and 1800 cm−1 .

homodimer and heterodimers. These changes can be explained by the difference of the discrete structures in the models and the rigid molecular environment associated with the crystal lattice. Conformational changes along single bonds can also be expected once a crystalline compound is transformed into its amorphous counterpart due to the loss of long range three dimensional order. Therefore, the optimized structures are suitable for the calculation of vibrational spectra. 3.2. Spectroscopic analysis The DFT prediction of the IR spectra was carried out on the optimized geometry of each structure. The resulting calculated and the experimental vibrational spectra are given in Figs. 3–6. Initially each of the pure compounds was characterized. The calculated monomer and homodimer structures of each drug were compared to the experimental crystalline and amorphous spectra. Furthermore, differences in the amorphous and crystalline spectra were analyzed. Then the experimental co-amorphous and calculated heterodimer spectra were investigated and compared to the individual amorphous drugs and homodimer predicitions. In general, the intensities for strong and weak bands in the calculations are close to those observed in the experimental spectra showing that the predictions are a good model. Furthermore, the experimental spectra of the amorphous form of each drug show band shifts and band broadening when compared to the individual crystalline spectrum. This observation is common with the crystalline to amorphous transformation of compounds (Heinz et al., 2009b). For both drugs in this study, specific changes are detectable in the vibrational regions of the hydrogen bonded carboxylic acid moiety (1700 cm−1 region) and the aromatic ring systems (1100–1400 cm−1 ). Therefore, the spectral region between 1000 and 1800 cm−1 was chosen for a detailed analysis. From the calculated spectra several bands in this region were matched to bands in the experimental spectra and are given in Tables 3 and 4 for

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

ARTICLE IN PRESS

G Model IJP-12600; No. of Pages 8

K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

5

Table 1 Dihedral angles for NAP from literature and after optimization. Line no.

Dihedral angles, Ф (◦ )

Kim and Song (1984)

N1 N2 N3 N4 N5 N6 N7 N8

Cn 1 Cn 1 Cn 3 Cn 3 On 1 On 1 On 2 On 2

−70.7 53.0 108.9 −127.4 88.7 −35.2 −88.8 147.3

Cn 2 Cn 9 Cn 10 Cn 2 Cn 9 Cn 11 Cn 2 Cn 9 Cn 10 Cn 2 Cn 9 Cn 11 Cn 10 Cn 9 Cn 2 Cn 10 Cn 9 Cn 11 Cn 10 Cn 9 Cn 2 Cn 10 Cn 9 Cn 11

Calculated B3-LYP 6-31G(d) Monomer

Homodimer

Heterodimer

−56.4 66.5 125.0 −112.2 89.1 −35.0 −89.5 146.4

−58.2 65.2 123.1 −113.4 84.9 −39.6 −94.2 141.4

−59.2 64.4 122.2 −114.3 85.2 −39.4 −93.8 141.7

Table 2 Dihedral angles for IND from literature and after optimization. Line no.

Dihedral angles, Ф (◦ )

Kistenmacher and Marsh (1972)

Calculated B3-LYP 6-31G(d) Monomer

Homodimer

Heterodimer

I1 I2 I3 I4 I5 I6 I7

Ci 2 Ni 1 Oi 1 Ci 2 Ci 3 Ci 3 Ci 5

−25.5 144.2 −39.3 99.9 146.7 −34.9 5.9

−29.6 154.4 −28.9 88.4 84.7 −93.8 0.1

−28.9 153.7 −29.6 99.1 93.5 −85.5 1.23

−29.3 153.9 −29.4 96.1 89.7 −89.0 0.8

Ni 1 Ci 10 Oi 1 Ci 10 Ci 11 Ci 12 Ci 10 Ci 11 Ci 12 Ci 3 Ci 18 Ci 19 Ci 18 Ci 19 Oi 3 Ci 18 Ci 19 Oi 4 Ci 6 Oi 2 Ci 20

Table 3 Selected calculated and experimental wavenumbers of NAP. Calculated monomer  (cm−1 )

Calculated homodimer  (cm−1 )

Calculated heterodimer  (cm−1 )

Experimental crystalline  (cm−1 )

Experimental amorphous  (cm−1 )

Experimental co-amorphous  (cm−1 )

Vibrational assignment

1135 1186

1227 1186

1224 1187

– 1193

1210 1194

– 1196

1220

1220

1218

1226

1228

1224

1256

1256

1256

1263

1263

1261

1602 1764

1601 1710

1603 1718

1603 1681

1604 1697

1605 1703

Out-of-plane On 2 Hn 2 bending, Cn 9H bending In-plane-naphthalene breathing, out-of-phase Cn 6 On 3 Cn 12 stretching, Naphthalene deformation, out-of-phase Cn 2 Cn 9 Cn 11 stretching, out-of-phase Cn 6 On 3 Cn 12 stretching Naphthalene deformation, out-of-phase Cn 6 On 3 Cn 12 stretching Cn 2 Cn 9 stretching, naphthalene deformation Asymmetric On 1 Cn 10 On 2 stretching (carboxylic acids out of phase in dimer)

NAP and IND, respectively. Since band assignment from quantum mechanical calculations for NAP (monomer) (Jubert et al., 2006) and IND (monomer and homodimer) (Strachan et al., 2007) can be found in literature, only a selection of bands are stated. The vibrational mode assignments are in good agreement with those mentioned in literature.

The calculated spectra for NAP homodimer and heterodimer are most different in the carboxylic acid vibration (Fig. 3). This is not surprising because a hydrogen bonding interaction is involved in the modeled dimer compared to the free carboxylic acid in the monomer. The carboxylic acid vibration in the homodimer has a 54 cm−1 lower wavenumber than in the monomer. The

Table 4 Selected calculated and experimental wavenumbers of IND. These modes and assignments are in good agreement with those reported by Strachan et al. (2007). Calculated monomer  (cm−1 )

Calculated homodimer  (cm−1 )

Calculated heterodimer  (cm−1 )

Experimental crystalline  (cm−1 )

Experimental amorphous  (cm−1 )

Experimental co-amorphous  (cm−1 )

Vibrational assignment

1111 1214

1271 1216

1269 1218

1291 1222

1289 1219

1288 1216

1246

1246

1246

1261

1259

1261

1298

1298

1298

1306

1314

1316

1604 1693 1779

1604 1692 1723

1603 1693 1718

1614 1689 1714

1608 1680 1708

1605 1680 1703

Out-of-plane Oi 3Hi 3 bending, Ci 12H2 bending Indole ring deformation, Ci 10 Ci 11 stretching, out-of-phase Ci 6 Oi 2 Ci 20 stretching, out-of-phase Ci 2 Ni 1 Ci 9 stretching Out-of-phase Ci 2 Ni 1 Ci 10 stretching and out-of-phase Ni 1 Ci 10 Ci 11 stretching, out-of-phase Ci 6 Oi 2 Ci 20 stretching, in-plane indole and chlorobenzene ring deformations, Ci 19 Oi 3 stretching (dimer) Out-of-phase and in-phase Ci 9 Ni 1 Ci 10 stretching, in-plane indole and chlorobenzene ring defomrations Indole ring deformation Ci 10 Oi 1 stretching Asymmetric Oi 3 Ci 19 Oi 4 stretching (carboxylic acids out-of-phase in dimer)

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8 6

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

Fig. 6. Calculated IR spectra of NAP homodimer (a), NAPIND heterodimer (b), and IND homodimer (c) between 1000 and 1800 cm−1 .

Fig. 4. IR spectra of experimental crystalline IND (a), experimental amorphous IND (b), calculated IND homodimer (c), and calculated IND monomer (d) between 1000 and 1800 cm−1 .

dimer vibration at 1710 cm−1 is much closer to the experimentally assigned bands for the hydrogen-bonded crystalline and amorphous NAP at 1681 cm−1 and 1697 cm−1 (Fig. 3 and Table 3), respectively. Therefore, the calculated homodimer represents a better model for comparison. The experimental spectra show additional bands at 1725 cm−1 and 1728 cm−1 in crystalline and amorphous NAP. These bands are also associated with the hydrogen bonded carbonyls but a likely to represent a different bonded conformation; hence the spectral shift. It is unlikely that they can be attributed to the symmetric carbonyl dimer vibration as this is

Fig. 5. Experimental IR spectra of amorphous NAP (a), co-amorphous NAPIND (b), and amorphous IND (c) between 1000 and 1800 cm−1 .

predicted to be at lower frequency and to have much less intensity than the observed features. When comparing the experimental crystalline and amorphous NAP spectra, the solid state transformation into an amorphous form results in peak shifts for the hydrogen bonded and carboxylic acid vibrations from 1681 cm−1 and 1725 cm−1 in the crystal to 1697 cm−1 and 1728 cm−1 . In addition, the intensities of these peaks differ greatly of each other. These differences strongly indicate a change in hydrogen bonding upon amorphization and rearrangement within the NAP molecules. Upon the NAP homodimer formation, a vibrational mode at 1227 cm−1 is predicted in the calculation (Fig. 3). This mode can be attributed to hydrogen bonded out-of-plane On 2 Hn 2 and Cn 9H bending (Table 3). The calculated mode for this vibration in the monomer is at 1135 cm−1 . Interestingly, the experimental spectrum of amorphous NAP also generates a new peak in this region at 1210 cm−1 compared to the crystalline spectrum. This further supports the assumption that amorphous NAP undergoes a change in hydrogen bonding with the formation of dimers. Similar to NAP, the hydrogen bonded IND dimer gave a better prediction of the experimental spectra than the IND monomer (Fig. 4 and Table 4). The calculated carboxylic acid vibration shifts from 1779 cm−1 in the monomer to 1723 cm−1 in the homodimer. In the experimental spectra for this particular vibration, a downward shift from 1714 cm−1 (crystalline) to 1708 cm−1 (amorphous) was detected. Furthermore, a shoulder at 1734 cm−1 appears in the amorphous spectrum which represents a fraction of carbonyl in a different hydrogen bonded environment and the benzoyl C O vibration shifts from 1689 cm−1 to 1680 cm−1 upon amorphization. Since the benzoyl C O is not involved any interactions in the model, the predicted vibrations here are nearly identical in the monomer and homodimer. The 1271 cm−1 vibration in the calculated IND homodimer corresponds to the hydrogen bonded Oi 3Hi 3 and Ci 18H2 out-of-plane bending similar to that in NAP. This mode can be matched with the experimental bands in crystalline (1291 cm−1 ) and amorphous (1289 cm−1 ) IND. In comparison, the calculation for the monomer results in a vibrational mode at 1111 cm−1 . This further supports the hypothesis by Taylor and Zografi (1997) and Strachan et al. (2007) that IND is forming homodimers in its amorphous form.

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

The experimental spectra for the individual amorphous drugs and the respective co-amorphous single form are given in Fig. 5. In general, peak shifts and changes in peak intensities for various bands can be detected when compared to the amorphous individual drugs. The main differences can be found for the carboxylic acid vibrations. In the co-amorphous mixture this peak appears, with increased intensity, at an intermediate wavenumber of 1703 cm−1 in between the vibrational modes for the carboxylic acid in amorphous NAP (1697 cm−1 ) and IND (1708 cm−1 ). In our previous study, this change in wavenumber was attributed to the formation of a heterodimer between both drugs (Löbmann et al., 2011). The quantum mechanical calculation predicted the same trend in the change of the vibrational modes (Fig. 6). The heterodimer carboxylic acid stretching (1718 cm−1 ) has an intermediate wavenumber to those predicted for the NAP (1710 cm−1 ) and IND homodimer (1723 cm−1 ). It is worth mentioning that the carboxylic acid stretching in the heterodimer resulted in a synchronized and single vibrational mode for both acid groups in NAP and IND. This finding strongly supports the assumption of a heterodimer formation. Furthermore, the calculated vibrations for the Oi 3Hi 3 (1271 cm−1 ) and On 2Hn 2 (1227 cm−1 ) bendings in the homodimers were predicted to shift slightly to lower wavenumbers in the heterodimer, i.e. to 1269 cm−1 and 1224 cm−1 , respectively. In the experimental spectra, the Oi 3Hi 3 band in amorphous IND (1289 cm−1 ) was found at 1288 cm−1 in the co-amorphous form. However, the change in wavenumber is below the spectral resolution. Unfortunately, the respective peak for the On 2Hn 2 bending in amorphous NAP (1210 cm−1 ) could not be matched with certainty to a peak in the co-amorphous spectrum. A broad peak at 1216 cm−1 with two shoulders at 1196 cm−1 and 1224 cm−1 appeared in the co-amorphous spectra. In general, major peak shifts were detectable for the experimental spectra in the area from 1194 to 1263 cm−1 . Since this area is also responsible for aromatic ring vibrations of both drugs (Tables 3 and 4), this also suggests a change in the molecular environment for the NAP naphthalene and IND indol and chlorobenzene rings upon the formation of a co-amorphous phase. A similar, but small, change for the indol (1216 cm−1 ) and naphthalene (1220 cm−1 ) vibrations was also detected in the calculations resulting in a vibrational mode at 1218 cm−1 for the heterodimer (Fig. 6). The experimental vibration for the benzoyl C O in amorphous IND (1680 cm−1 ) is the same in the co-amorphous spectrum (Fig. 5). The observation that the band has not shifted suggests that the C O is not involved in any hydrogen bonding. The calculations are consistent with this; there is no predicted linkage involving the C O group and the vibrational modes for the benzoyl C O in the IND homodimer and heterodimer are at similar predicted wavenumbers. In our previous work, IR spectroscopy was used to identify interactions between IND and NAP. It was suggested that homodimers would form in the amorphous state of the single compounds. In contrast, when NAP and IND are quench-cooled together, a heterodimer is formed in the single phase co-amorphous mixture. The results presented here further support this hypothesis. It can be discussed whether the observed results are due to the formation of a heterodimer or simply due to a mixture of homodimers, as the presence of a mixture of IND and NAP homodimers will give spectra similar to the heterodimer. The calculation of the energetics upon the formation of the various homoand hetero-dimer forms is inconclusive in this point as it shows that dimer formation is favorable in all cases – both homodimer and heterodimer are energetically favorable by about 20 kcal/mol. The energetics of formation of the dimer were determined by comparing the “sum of electronic and thermal free energies” in

7

the frequency output file for the respective dimer with that of the reactant monomers (Ochterski, 2000). However, although the experimental data depicted in Fig. 5 show many similarities, the experimental data for the co-amorphous mixture is not a simple linear combination of spectra of the homodimers. This is supportive of the conclusion that the spectrum is indeed due to the formation of a heterodimer. The formation of a heterodimer in a single phase co-amorphous mixture also suggests the possibility of a co-crystal formation in the same molecular arrangement upon recrystallization. However, this was not observed in the case of co-amorphous NAP and IND. Upon recrystallization, X-ray powder diffraction revealed the presence of both crystalline compounds indicating the inability of both drugs to crystallise into a co-crystal and the separation back into single crystalline compounds (data not shown). This finding suggests, that even if the formation of a co-crystal is not successful, there is still the possibility of creating a single phase co-amorphous mixture of two small molecular compounds upon molecular interactions. 4. Conclusion In this study, quantum mechanical calculations were successfully applied to investigate the structure and IR spectra of NAP and IND as individual amorphous drugs and in a co-amorphous (1:1 molar ratio) combination after quench cooling. The analysis of the vibrational modes of the individual drugs suggested the formation of homodimers in the respective single amorphous solid state forms. Furthermore, characterization of the experimental coamorphous IR spectrum and the quantum mechanical calculation of the heterodimer provided insight into the molecular structure in co-amorphous NAP–IND. In particular, the formation of a heterodimer between NAP and IND when quench cooled together could be confirmed by the use of density functional theory calculations. This study highlights the usefulness of quantum mechanical calculations as a valuable tool to improve the understanding of structural properties of amorphous and most importantly coamorphous systems. Acknowledgements The authors would like to thank Ms. A.B.S. Elliot (Dept. of Chemistry, University of Otago) for her help in conducting the DFT calculations. The Academy of Finland, Magnus Ehrnrooth Foundation and Saastamoinen Foundation are also acknowledged for funding (RL). References Aaltonen, J., Rades, T., 2009. Towards physico-relevant dissolution testing: the importance of solid-state analysis in dissolution. Dissolution Technol. 16, 47–54. Ali, H.R.H., Edwards, H.G.M., Kendrick, J., Munshi, T., Scowen, I.J., 2007. Vibrational spectroscopic study of budesonide. J. Raman Spectrosc. 38, 903–908. Ali, H.R.H., Edwards, H.G.M., Kendrick, J., Munshi, T., Scowen, I.J., 2010. An experimental and computational study on the epimeric contribution to the infrared spectrum of budesonide. Drug Test. Anal. 2, 447–451. Allesø, M., Chieng, N., Rehder, S., Rantanen, J., Rades, T., Aaltonen, J., 2009. Enhanced dissolution rate and synchronized release of drugs in binary systems through formulation: amorphous naproxen-cimetidine mixtures prepared by mechanical activation. J. Control. Release 136, 45–53. Ayala, A.P., Siesler, H.W., Boese, R., Hoffmann, G.G., Polla, G.I., Vega, D.R., 2006. Solid state characterization of olanzapine polymorphs using vibrational spectroscopy. Int. J. Pharm. 326, 69–79. Bar, I., Bernstein, J., 1985. Conformational polymorphism VI: the crystal and molecular structures of form II, form III, and form V of 4-amino-N-2pyridinylbenzenesulfonamide (sulfapyridine). J. Pharm. Sci. 74, 255–263. Bernstein, J., Hagler, A.T., 1978. Conformational polymorphism. The influence of crystal structure on molecular conformation. J. Am. Chem. Soc. 100, 673–681. Borba, A., Gómez-Zavaglia, A., Fausto, R., 2009. Molecular structure, infrared spectra, and photochemistry of isoniazid under cryogenic conditions. J Phys. Chem. A 113, 9220–9230.

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016

G Model IJP-12600; No. of Pages 8 8

ARTICLE IN PRESS K. Löbmann et al. / International Journal of Pharmaceutics xxx (2012) xxx–xxx

Chesalov, Y.A., Baltakhinov, V.P., Drebushchak, T.N., Boldyreva, E.V., Chukanov, N.V., Drebushchak, V.A., 2008. FT-IR and FT-Raman spectra of five polymorphs of chlorpropamide. Experimental study and ab initio calculations. J. Mol. Struct. 891, 75–86. Chieng, N., Aaltonen, J., Saville, D., Rades, T., 2009. Physical characterization and stability of amorphous indomethacin and ranitidine hydrochloride binary systems prepared by mechanical activation. Eur. J. Pharm. Biopharm. 71, 47–54. Chieng, N., Rades, T., Aaltonen, J., 2011. An overview of recent studies on the analysis of pharmaceutical polymorphs. J. Pharm. Biomed. Anal. 55, 618–644. Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., Cheeseman, J.R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G.A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H.P., Izmaylov, A.F., Bloino, J.Z., Zheng, G., Sonnenberg, J.L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, J.J.A., Peralta, J.E., Ogliaro, F., Bearpark, M., Heyd, J.J., Brothers, E., Kudin, K.N., Staroverov, V.N., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J.C., Iyengar, S.S., Tomasi, J., Cossi, M., Rega, N., Millam, N.J., Klene, M., Knox, J.E., Cross, J.B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R.E., Yazyev, O., Austin, A.J., Cammi, R., Pomelli, C., Ochterski, J.W., Martin, R.L., Morokuma, K., Zakrzewski, V.G., Voth, G.A., Salvador, P., Dannenberg, J.J., Dapprich, S., Daniels, A.D., Farkas, Ö., Foresman, J.B., Ortiz, J.V., Cioslowski, J., Fox, D.J., 2009. Gaussian 09, Revision A.1. Gaussian, Inc., Wallingford, CT. Gordon, K.C., McGoverin, C.M., Strachan, C.J., Rades, T., 2007. The use of quantum chemistry in pharmaceutical research as illustrated by case studies of indometacin and carbamazepine. J. Pharm. Pharmacol. 59, 271–277. Hancock, B.C., Zografi, G., 1997. Characteristics and significance of the amorphous state in pharmaceutical systems. J. Pharm. Sci. 86, 1–12. Heinz, A., Gordon, K.C., McGoverin, C.M., Rades, T., Strachan, C.J., 2009a. Understanding the solid-state forms of fenofibrate – a spectroscopic and computational study. Eur. J. Pharm. Biopharm. 71, 100–108. Heinz, A., Strachan, C.J., Gordon, K.C., Rades, T., 2009b. Analysis of solid-state transformations of pharmaceutical compounds using vibrational spectroscopy. J. Pharm. Pharmacol. 61, 971–988. Hu, Y., Erxleben, A., Ryder, A.G., McArdle, P., 2010. Quantitative analysis of sulfathiazole polymorphs in ternary mixtures by attenuated total reflectance infrared, near-infrared and Raman spectroscopy. J. Pharm. Biomed. Anal. 53, 412–420. Iliescu, T., Baia, M., Kiefer, W., 2004. FT-Raman, surface-enhanced Raman spectroscopy and theoretical investigations of diclofenac sodium. Chem. Phys. 298, 167–174. Jubert, A., Legarto, M.L., Massa, N.E., Tévez, L.L., Okulik, N.B., 2006. Vibrational and theoretical studies of non-steroidal anti-inflammatory drugs Ibuprofen [2-(4-isobutylphenyl)propionic acid]; naproxen [6-methoxy-[alpha]-methyl-2naphthalene acetic acid] and tolmetin acids [1-methyl-5-(4-methylbenzoyl)1H-pyrrole-2-acetic acid]. J. Mol. Struct. 783, 34–51. Kim, Y., Song, H., 1984. The molecular structure of (+)-6-methoxy-␣-methyl-2naphthaleneacetic acid determined by X-ray method. Arch. Pharmacal Res. 7, 137–139. Kim, Y., Song, H., Park, I., 1987. Refinement of the structure of naproxen, (+)-6-methoxy-␣-methyl-2-naphthaleneacetic acid. Arch. Pharmacal Res. 10, 232–238.

Kistenmacher, T.J., Marsh, R.E., 1972. Crystal and molecular structure of an antiinflammatory agent, indomethacin, 1-(p-chlorobenzoyl)-5-methoxy-2methylindole-3-acetic acid. J. Am. Chem. Soc. 94, 1340–1345. Löbmann, K., Laitinen, R., Grohganz, H., Gordon, K.C., Strachan, C., Rades, T., 2011. Coamorphous drug systems: enhanced physical stability and dissolution rate of indomethacin and naproxen. Mol. Pharma. 8, 1919–1928. Löbmann, K., Strachan, C., Grohganz, H., Rades, T., Korhonen, O., Laitinen, R., 2012. Co-amorphous simvastatin and glipizide combinations show improved physical stability without evidence of intermolecular interactions. Eur. J. Pharm. Biopharm. 81 (1), 159–169. Mishra, S., Chaturvedi, D., Tandon, P., Gupta, V.P., Ayala, A.P., Honorato, S.B., Siesler, H.W., 2008. Molecular structure and vibrational spectroscopic investigation of secnidazole using density functional theory. J. Phys. Chem. A 113, 273–281. Müllertz, A., Ogbonna, A., Ren, S., Rades, T., 2010. New perspectives on lipid and surfactant based drug delivery systems for oral delivery of poorly soluble drugs. J. Pharm. Pharmacol. 62, 1622–1636. Nolasco, M.M., Amado, A.M., Ribeiro-Claro, P.J.A., 2006. Computationally-assisted approach to the vibrational spectra of molecular crystals: study of hydrogenbonding and pseudo-polymorphism. ChemPhysChem 7, 2150–2161. O’Boyle, N.M., Tenderholt, A.L., Langner, K.M., 2008. cclib: a library for packageindependent computational chemistry algorithms. J. Comput. Chem. 29, 839–845. Ochterski, J.W., 2000. Thermochemistry in Gaussian. http://www.gaussian.com/ g whitepap/thermo.htm (accessed 25.02.2012). Sagdinc, S., Kandemirli, F., Bayari, S.H., 2007. Ab initio and density functional computations of the vibrational spectrum, molecular geometry and some molecular properties of the antidepressant drug sertraline (Zoloft) hydrochloride. Spectrochim. Acta Part A: Mol. Biomol. Spectrosc. 66, 405–412. Scott, A.P., Radom, L., 1996. Harmonic vibrational frequencies: an evaluation of Hartree–Fock, Møller–Plesset, quadratic configuration interaction, density functional theory, and semiempirical scale factors. J. Phys. Chem. 100, 16502–16513. Srivastava, A., Mishra, S., Tandon, P., Patel, S., Ayala, A.P., Bansal, A.K., Siesler, H.W., 2010. Molecular structure and vibrational spectroscopic analysis of an antiplatelet drug; clopidogrel hydrogen sulphate (form 2) – a combined experimental and quantum chemical approach. J. Mol. Struct. 964, 88–96. Strachan, C.J., Howell, S.L., Rades, T., Gordon, K.C., 2004. A theoretical and spectroscopic study of carbamazepine polymorphs. J. Raman Spectrosc. 35, 401–408. Strachan, C.J., Rades, T., Gordon, K.C., 2007. A theoretical and spectroscopic study of ␥-crystalline and amorphous indometacin. J. Pharm. Pharmacol. 59, 261–269. Taylor, L.S., Zografi, G., 1997. Spectroscopic characterization of interactions between pvp and indomethacin in amorphous molecular dispersions. Pharm. Res. 14, 1691–1698. Vueba, M.L., Pina, M.E., Batista de Carvalho, L.A.E., 2008. Conformational stability of ibuprofen: assessed by DFT calculations and optical vibrational spectroscopy. J. Pharm. Sci. 97, 845–859. Vueba, M.L., Pina, M.E., Veiga, F., Sousa, J.J., de Carvalho, L.A.E.B., 2006. Conformational study of ketoprofen by combined DFT calculations and Raman spectroscopy. Int. J. Pharm. 307, 56–65.

Please cite this article in press as: Löbmann, K., et al., A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin. Int J Pharmaceut (2012), http://dx.doi.org/10.1016/j.ijpharm.2012.05.016