- Email: [email protected]
Establishment of cocrystal cocktail grinding method for rational screening of pharmaceutical cocrystals
International Journal of Pharmaceutics 437 (2012) 162–171
Contents lists available at SciVerse ScienceDirect
International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm
Establishment of cocrystal cocktail grinding method for rational screening of pharmaceutical cocrystals Katsuhiko Yamamoto a,∗ , Shunichirou Tsutsumi b , Yukihiro Ikeda a a Drug Metabolism and Pharmacokinetics Research Laboratories, Pharmaceutical Research Division, Takeda Pharmaceutical Company Limited, 2-26-1, Muraoka-Higashi, Fujisawa, Kanagawa, 251-8555, Japan b Pharmaceutical Technology Research and Development Laboratories, CMC Center, Takeda Pharmaceutical Company Limited, 2-17-85, Juso-Hommachi, Yodogawa-ku, Osaka, 532-8686, Japan
a r t i c l e
i n f o
Article history: Received 20 March 2012 Received in revised form 2 July 2012 Accepted 22 July 2012 Available online 31 July 2012 Keywords: Cocrystal Grinding Drug substance Amorphous Crystallinity Physicochemistry
a b s t r a c t Cocrystals (CCs) used in the pharmaceutical industry are defined as complex crystals formed by reaction between an API and a cocrystal former (CCF); unlike salts, CCs do not show proton transfer. Recently, pharmaceutical CCs have been used to improve the drug-likeness of APIs, such as solubility and stability. Grinding is more effective for CC synthesis than crystallization from solution because in the former case, the API can predominantly interact with the CCF without being affected by solvents. However, this method is tedious because the API is ground with only one CCF at a time. We developed a cocktail cocrystal grinding (CCG) method, in which a mixture of CCFs having the same functional group was used. No false negatives/positives were observed in CCG when carbamazepine was used as the model compound. This method could be used to obtain CCs of piroxicam and spironolactone. False negatives were observed for only one compound from among three model compounds, indicating that CCG facilitates efficient CC detection and that it has higher throughput than does the conventional method. Further, CCG is fast and suitable for rational CC screening, and it helps identify the partial structure of CCFs that forms synthons with an API. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Recently, cocrystals (CCs) of drug substances have been studied extensively with the aim of improving the physicochemical properties of drug substances: solubility, stability, hygroscopicity, etc. (Vishweshwar et al., 2006; Shan and Zaworotko, 2008; Schultheiss and Newman, 2009; Lee et al., 2011). A CC is defined as a complex crystal whose lattice includes more than two chemicals that are in the solid state at room temperature (Almarsson and Zaworotko, 2004). Although salts can be classified as CCs, they show proton transfer, while the CCs fabricated in this study do not. In a CC lattice, a host molecule (drug) interacts with guest molecules, cocrystal formers (CCF), via hydrogen bonding and other weak interactions (Aakeröy et al., 2001; Remenar et al., 2003; Limwikrant et al., 2009). Various methods for crystallizing CCs have been investigated, such as crystallization from solution (Ainouza et al., 2009;
Abbreviations: CC, cocrystal; CCF, cocrystal former; API, active pharmaceutical ingredient; PXRD, powder X-ray diffraction; TG, thermogravimetry; DSC, differential scanning calorimetry. ∗ Corresponding author. Tel.: +81 466 32 2850; fax: +81 466 29 4432. E-mail addresses: Yamamoto [email protected], [email protected] (K. Yamamoto). 0378-5173/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpharm.2012.07.038
Rodriguez-Hornedo et al., 2005; Basavoju et al., 2008), crystallization from slurry (Hickey et al., 2007; Takata et al., 2008; Kojima et al., 2010), and melt recrystallization (Seefeldt et al., 2007). Among these, the most popular method is crystallization from solution, which includes cooling, evaporation, and addition of an antisolvent. However, it is often difficult to crystallize CCs from solution because of the following reasons: (1) the large difference between the solubilities of the drug and the CCF (Chiarella et al., 2007); (2) the strong interaction with solvents; (3) and nonattainment of the critical activity to form CCs (Schartman, 2009). Grinding is one of the important methods used to crystallize CCs (Trask et al., 2006; Fukami et al., 2006; Karki et al., 2007; Jayasankar et al., 2006; Friˇscˇ ic´ and Jones, 2009). A schematic illustration of crystallization by grinding is shown in Fig. 1. This method has advantages over crystallization from solution because in the former case, interactions between the drug and the CCFs occur in the solid state. Some of these advantages are as follows: (1) the interaction between the solvent and the drug (or CCF) does not disturb the interaction between the drug and the CCF; (2) since there is no solubility limit in the system, the CCF can be added to the system until the activity of the drug is sufficient for interaction with the CCF and formation of CCs with the API. However, grinding is time-consuming and tedious because each of the CCFs being used needs to be ground with the drug individually. In this study, we
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
163
2.3. Powder X-ray diffraction (PXRD) About 1 mg of the powder was placed on a slide glass and analyzed with a powder X-ray diffractometer, Bruker D8 Discover HTS with GADDS (Madison, WI, USA). The instrument parameters were as follows: X-ray source power, 40 kV/40 mA; collimator radius, 0.3 mm; detector distance, 25 cm; exposure time, 120 s; and detection angle range (2), 5–40◦ .
Fig. 1. Schematic of CC formation by grinding.
2.4. Thermal analysis (TG/DSC) established cocktail cocrystal grinding (CCG), which is a rational method for CC screening. This novel method enables four CCFs to be ground with API simultaneously and hence is more advantageous than the conventional method. The CCFs in the mixture have the identical chemical moieties that interact with the chemical moiety of the drug molecule and form synthons (the minimum unit of interaction between a drug and a CCF). Therefore, the CCG method decreases time, labor, and the amount of API that is to be ground with multiple CCFs simultaneously; further, it enables us to identify the partial structure of the CCFs that preferentially forms synthon with the drug. 2. Materials and methods 2.1. Materials Carbamazepine, piroxicam was obtained from Wako Pure Chemical Industries (Osaka, Japan). Spironolactone, lactamide, ethylmaltol and N-methyl-d-glucamine were obtained from Tokyo Kasei Kogyo (Tokyo, Japan). Glycoamide were obtained from Sigma–Aldrich (St. Louis, MO, USA). The other cocrystal formers were obtained from Wako Pure Chemical Industries (Osaka, Japan). All solvents were purchased from Wako Pure Chemical Industries (Osaka, Japan). 2.2. Grinding operation About 20 mg of powder containing an API and CCFs and a 7-mm stainless-steel ball were placed in a 1.5-mL stainless-steel cell from Retsch (Haan, Germany). Then, 20 L of n-heptane was added to the cell. The cell was shaken in a Retsch MM301 mixer mill (Haan, Germany) at 28.5 rps for 30 min.
About 2 mg of the powder was sampled and placed in an open aluminum crucible and analyzed with a Mettler-Toredo TGA/DSC1 simultaneous TG/DSC analyzer (Greifensee, Switzerland). The instrument parameters were as follows: ramp rate, 5 ◦ C/min; temperature range, 25–300 ◦ C; atmosphere, nitrogen gas; and gas flow rate, 40 mL/min. 3. Results and discussion 3.1. Design and properties of CCF mixture Five CCF mixtures were designed, as shown in Table 1. Each CCF mixture, except the Others group, contains four equimolar CCFs having the same chemical moiety to form synthons. The five CCF mixtures were prepared with various specifications. First, the safety and feasibility of using these CCFs was considered, because these CCFs are excipients, additives, or counter-ions of pharmaceutical salts used in commercially available drug substances and products. Second, these CCFs have been referred to in publications on CCs. CCs synthesized from organosulfonic acids have often been reported, however organosulfonic acids have not been adopted in the present study because they are highly deliquescent and the alcoholic esters of these acids may be genotoxic. Third, CCFs with high aqueous solubility were selected. Some CCs obtained from an API and such CCFs improved the aqueous solubility of the API itself (McNamara et al., 2006; Tsutsumi et al., 2011). During the grinding process, two CCFs in the mixture can react to form CCs. Hence, the CCFs were gently micronized and mixed in a mortar and pestle to avoid mechanochemical changes in the crystallinity and the crystal form. Grinding of the CCF mixtures was carried out solely to identify whether CCs can be formed from
Table 1 Physicochemical properties of CCF mixtures and individual CCFs. CCFs Mixture
CCF Name
Molecular weight (Da)
H-Acceptor
H-Donor
Melting point (◦ C)
Acid 1 (Di-, tri-carboxylic acid)
Citric acid Fumaric acid Succinic acid l-Tartaric acid
192 116 118 151
7 4 4 6
4 2 2 4
135 210 (sublimation) 185 168
Acid 2 (Aromatic, large carboxylic acid)
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
122 138 154 132
2 3 4 4
1 2 3 2
121 159 199 (degradation) 95
Amide (amide)
Nicotinamide Benzamide Glycolamide Lactamide
122 121 75 89
3 2 3 3
1 1 2 2
129 128 122 120
Amine (amine)
Tromethamine Megluimine l-Arginine l-lysine
121 195 174 146
4 6 6 4
4 6 5 3
167 129 222 215
Others (other CCFs)
Urea Ethylmaltol Saccharin Stearic acid
60 140 183 284
3 3 4 2
2 1 1 1
132 (sublimation) 85 (sublimation) 226 (sublimation) 67 (sublimation)
164
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 2. PXRD patterns of CCF mixtures before and after grinding. Inset table shows a detailed annotation of each pattern.
the CCF mixture or not. The PXRD patterns of the CCF mixtures before and after grinding are shown in Figs. 2a (Acid 1 and Acid 2), Fig. 2b (Amide and Amine), and Fig. 2c (Others). No new diffraction peaks were identified in the PXRD patterns of Acid 2 and Amide after grinding. On the other hand, minor diffraction peaks appeared in the PXRD patterns of Acid 1, Amine, and Others after grinding. These new peaks might be attributed to CCs formed by the reaction between CCFs, but the amount of CCs formed was very less. Hence, contamination by these CCs would not affect the formation of CCs between the API and the CCFs, and the resulting CCs could be detected by PXRD. 3.2. Trial CCG with model API (1) – carbamazepine (CBZ) chose carbamazepine (5H-dibenz[b,f]azepine-NWe carboxamide; Fig. 3) as one of the model compounds for CCG. Carbamazepine (CBZ) is a well known and widely used anticonvulsant marketed under the name Tegretol® . Since CBZ is a
Fig. 3. Chemical structure of CBZ.
neutral compound with low solubility in aqueous media, several studies have been carried out to improve its solubility. Homoor heterosynthons can be formed from the amide group of CBZ and the partial structures of the CCFs in CCs (Schartman, 2009). Since the CCs of CBZ were extensively investigated using various crystallization methods (Childs et al., 2008), CBZ is suitable for the model compound of CC. The PXRD patterns for CBZ–CCF mixtures after grinding are depicted in Fig. 4a (CBZ-Acid 1 and CBZ-Acid 2), Fig. 4b (CBZ-Amide and CBZ-Amine), and Fig. 4c (CBZ-Others). Several new diffraction peaks appeared in the PXRD patterns of CBZ-Acid 1, CBZ-Acid 2, CBZ-Amide, and CBZ-Others after grinding. No new diffraction peaks were identified in the PXRD pattern of CBZ-Amine. This indicated CC formation between CBZ and the CCFs in Acid 1, Acid 2, Amide, and Others. We ground CBZ with individual CCFs in the CCF mixture and closely scrutinized the grinding process, to confirm CC formation and identify the CCF that formed CCs with CBZ. If some of the CCFs in the CCF mixture form CCs with CBZ, the mixture would also form the same CCs with CBZ. Acid 1, Acid 2, Amide, and Others were thought to contain CCFs that could form CCs with CBZ. Fig. 5a–d show the PXRD patterns of CBZ with the individual CCFs in Acid 1, Acid 2, Amide, and Others after grinding, respectively. In the case of Acid 1 group, a few unique peaks were identified in the patterns of CBZ-fumaric acid, CBZ-succinic acid, and CBZ-ltartaric acid. These data indicated that CBZ forms CCs with fumaric acid, succinic acid and l-tartaric acid. Similarly, in the case of the Acid 2 group, a few unique peaks were identified in the PXRD patterns of CBZ-benzoic acid, CBZ-salicylic acid, CBZ-gentisic acid, and CBZ-l-glutaric acid. These data indicated that CBZ also formed CCs
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
165
Fig. 4. PXRD patterns of CBZ–CCF mixtures after grinding. Inset table shows a detailed annotation of each pattern.
with benzoic acid, salicylic acid, gentisic acid, and l-glutaric acid. The carboxamide group of CBZ easily forms a heterosynthon with a carboxylic acid that links an aromatic ring. Grinding CBZ with each CCF in Amide resulted in the formation of CBZ-nicotinamide and CBZ-benzamide CCs. Aromatic amides
preferentially form synthons with the carboxamide of CBZ, but aliphatic amides do not form CCs with CBZ. Grinding CBZ with individual CCFs in Others resulted in the formation of CBZ-urea and CBZ-saccharin CCs. CBZ-urea is a novel CC of CBZ.
Table 2 Results summary of CCG method using three model compounds, CBZ, PRX and SPI. Carbamazepine
Piroxicam
Spironolactone
CCF mixture
Individual CCF
CCF mixture
Individual CCF
CCF mixture
Individual CCF
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amine
Tromethamine Meglumine l-Arginine l-lysine
Amine
Tromethamine Meglumine l-Arginine l-lysine
Amine
Tromethamine Meglumine l-Arginine l-lysine
Others
Urea Ethylmaltol Saccarin Stearic acid
Others
Urea Ethylmaltol Saccarin Stearic acid
Others
Urea Ethylmaltol Saccarin Stearic acid
Bold: hit and reported CC, bold and underline: novel CC, normal: no CC.
166
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 5. PXRD patterns of CBZ-individual CCFs in CCF mixtures after grinding. Inset table shows a detailed annotation of each pattern.
To confirm if any false negatives are observed in this method, we carried out grinding of each CCF in Amine with CBZ. Fig. 5e shows the PXRD patterns of the individual CCFs in Amine: tromethamine, meglumine, l-lysine, and l-arginine. No new peak was identified in the patterns, indicating that CBZ did not form CCs with these amines. From these observations, it was clear that the carboxamide of CBZ does not form heterosynthons with amines. The CCG method does not give false negatives, hence, new CCs of APIs can be identified accurately. In addition, there were no false positives in the trial performed using CBZ. 3.3. Trial CCG with model APIs (2) – piroxicam (PRX) and spironolactone (SPI) To verify the adaptability of CCG method, we tested a nonsteroidal anti-inflammatory drug, piroxicam (PRX, 4-hydroxy2-methyl-N-pyridin-2-yl-2H-1,2-benzothiazine-3-carboxamide 1,1-dioxide) and spironolactone (SPI, 7␣-acetylthio-3-oxo-17␣pregn-4-ene-21,17-carbolactone), which is a diuretic. The chemical
structures of PRX and SPI are shown in Fig. 6. PRX has a carbonyl group, which can form synthons with the partial structures of CCFs. Several CCs of piroxicam have been identified in a previous study (Childs and Hardcastle, 2007). SPI is a neutral compound, and it has some carbonyl groups that form synthons with CCF. Since the aqueous solubility of SPI is low, the improvement of the solubility is required to increase the bioavailability of this API. We assumed that CCs of PRX and SPI were detected by CCG, as CCs of CBZ were clearly identified by this method. Table 2 summarizes the results of CCG trials using three model compounds (CBZ, PRX and SPI). On the PRX trial, new diffraction peaks did not appear in the PXRD patterns of PRX-Acid 1, PRX-Amide, and PRX-Amine, while some new diffraction peaks were identified in the PXRD patterns of PRX-Acid 2 and PRX-Others. These observations confirmed that PRX formed CCs with some of the CCFs in Acid 2 and Others. To confirm the CC formation between PRX and the CCFs and evaluate false positives/negatives in the proposed method, we ground PRX with individual CCFs. Some new diffraction peaks were observed in the PXRD patterns of PRX-citric acid and PRX-succinic acid (from
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
167
Fig. 6. Chemical structure of PRX and SPI.
Acid 1); PRX-benzoic acid, PRX-salicylic acid, PRX-gentisic acid, and PRX-glutaric acid (from Acid 2); and PRX-ethyl maltol (from Others). These results suggested that PRX formed CCs with these CCFs. The formation of PRX-citric acid and PRX-succinic acid CCs was considered as false negatives because PRX did not form CCs with the Acid 1 CCF mixture. PRX-ethyl maltol is a novel CC of PRX. The representative PXRD patterns of PRX with CCF mixtures, Others, and individual CCFs in Others after grinding are depicted in Fig. 7. Next, we ground SPI with CCF mixtures. New diffraction peaks appeared only in the PXRD patterns of PRX-Acid 2. In the case of grinding with the individual CCFs in the CCF mixtures, new diffraction peaks were also observed in the PXRD patterns of SPIbenzoic acid, SPI-salicylic acid and SPI-gentisic acid, which belong to Acid 2 mixture. Fig. 8 depicts the representative PXRD patterns of SPI with CCF mixtures, Acid 2, and individual CCFs in Acid2 after grinding. Therefore, SPI-benzoic acid CC, SPI-salicylic acid CC and SPI-gentisic acid CC were identified. These CCs are novel CCs of SPI. Aromatic carboxylic acids preferentially form synthons with SPI. No false positive/negative was identified. Although SPI-saccharin CCs were reported (Takata et al., 2010), these CCs were not identified in our experiments. In that study, SPI-saccharin CCs was primarily crystallized as hydrates. Since grinding was performed under anhydrous conditions in this study, these CCs were not crystallized.
3.4. Scaleup and characterization of identified CCs We conducted scaleup and characterization of the identified CCs of CBZ, PRX and SPI in order to validate the CC formation and to evaluate physicochemical properties of the CCs. CBZ-urea CC, PRXethyl maltol CC, and SPI-gentisic acid CC were selected from among the CCs formed by CCG. Scaleup was conducted by simply increasing the cell size (1.5–5.0 mL) and sample amount (20–100 mg). The PXRD patterns of the CCs during grinding are shown in Fig. 9a (CBZurea CC), Fig. 9b (PRX-ethyl maltol CC), and Fig. 9c (SPI-gentisic acid CC). The diffraction peaks due to the starting material remained in the PXRD patterns of the ground samples even after 30 min, however a very small amount of the starting materials remained in the ground samples after 60 min. The observations suggested that a longer grinding time was necessary for the scaleup, since crystallization by grinding needs sufficient mechanical force for reaction, amorphization, and mixing (Chieng et al., 2009). The TG/DSC curves of the CCs are depicted in Fig. 10. The melting points of the CBZ-urea and SPI-gentisic acid CCs were different from those of the APIs and CCFs. PRX-ethyl maltol CCs showed a large weight loss with an endothermic event in the temperature range 100–120 ◦ C. The weight loss (about 25%) nearly corresponded to the theoretical amount of ethyl maltol when the PRX/ethyl maltol stoichiometry in PRX-ethyl maltol CC was 1:1. The weight loss was thought to
Fig. 7. PXRD patterns of PRX after grinding. (a) PRX-Others mixture (b) PRX-individual CCFs in Others mixture.
168
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 8. PXRD patterns of SPI after grinding. (a) SPI-Acid 2 mixture (b) SPI-individual CCFs in Acid 2 mixture.
be due to the dissociation of ethyl maltol from the CCs, similar to dehydration of hydrate crystals on heating. After dissociation, the PRX-ethyl maltol CCs were converted to PRX; this conversion was confirmed by the endothermic peak appearing at 201 ◦ C, which is the melting point of PRX form I. 3.5. Effects of dropped solvents into grinding N-Heptane was used as the dropped solvent on the grinding procedure in the CCG method. The effect and role of the dropped solvent was properly examined using CBZ-urea CC, which was
detected in the CCG method and crystallized again in the scaleup. We operated solvent drop grinding of CBZ and urea with the same amount of n-heptane, water and acetone, and without solvents, neat grinding. Water and acetone are good solvent for urea and CBZ, respectively. After 30 min grinding, CBZ-urea CC was detected in all the samples, however, CBZ form III remained in the samples of water, acetone and neat grinding. The residual CBZ was not observed in the sample of n-heptane. When the neat grinding was continued, only CBZ-urea CC was identified in the ground sample after 60 min. It is thought that solvents accelerate the reaction by raising the homogeneity and flowability of the powder mixture;
Fig. 9. PXRD patterns of novel CCs on scaleup experiments. Inset table shows a detailed annotation of each pattern.
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
169
Fig. 10. TG (a, c, e) and DSC (b, d, f) curves of APIs, CCFs and novel CCs. (a and b) CBZ; (c and d) PRX, (e and f) SPI.
however, good solvents disturb the reaction between API and CCF due to the dissolving ability, i.e. interaction with API or CCF. Therefore, n-heptane, which is a poor solvent for most of APIs and CCFs, is appropriate to grinding CC crystallization. Especially, because the CCG method uses the CCF mixtures that include various CCFs having each diverse solubility, n-heptane is very beneficial as the dropped solvent. 3.6. Rationality of CCG The rationality of CC screening by CCG should be discussed in comparison with the conventional single (one-to-one) grinding method. In the case of CC screening with the conventional method, 20 grinding experiments were required to identify one CC from among the CCs formed between the API and the 20 selected CCFs. On the other hand, when using CCG, only nine experiments were required to identify one CC; these experiments included five experiments with CCF mixtures and four experiments with the individual
CCFs in the mixtures. A comparison of the conventional method with CCG is illustrated in Fig. 11. The workload for CC screening in CCG is about 50% of that in the single method. Additionally, in CCG, only small amounts of APIs are required. The sample amount of API is often limited in early stage of preclinical study. The CCG method is relevant to CC screening in drug discovery. The reliability of the CCG method against the single method should also be considered. From among the three model compounds, false negatives were identified in only two cases: PRX-citric acid CCs and PRX-succinic acid CCs. On the contrary, no false positive was observed. The following issues need to be addressed in the case of CCG: (1) CC formation between a CCF and the API may compete with CC formation between some CCFs; (2) ternary CCs such as API-CCF 1CCF 2 may be formed by the interaction of two different CCs formed from the API and a CCF. The former competition is thought to block CC formation between the API and a single CCF, which may lead to a false negative. Examples of this blocking could be the formation of
170
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 11. Comparison of conventional method and CCG method.
PRX-citric acid CC and PRX-succinic acid CC in the trial performed on PRX. The latter ternary complex may produce a false positive. In the CCG method, when CC formation is identified on grinding an API with the CCF mixture, CCs are invariably identified on grinding with the individual CCFs. False positives were not observed in this study. Furthermore, the pros and cons of the CCG method are considered. Comparing with single grinding, multiple CCFs compete against the interaction with API molecule in the CCG method. The CCG method is easy to detect CCs that have physically stronger interaction with API. The strength of the interaction was not known in single grinding method. However, the CCG method might miss the CCs that have the weak interaction. Such CCs are able to be favorably detected by single grinding method. In fact, PRXcitric acid CC and PRX-succinic acid CC were not identified in the CCG method trials. From the point of the structure of CC, single grinding might presume the stoichiometry of a API and a CCF in a CC. On the other hand, the CCG method cannot anticipate it because the detected CCs in the CCG method are the mixture of CCs. Next, the comparison with crystallization from solution is discussed. If the CC screening by crystallization from solution using CCF mixture is performed, the detected CCs are thought to be easily crystallized CCs, not the CCs that have the physically strong interaction as detected in the CCG method. The crystallization ability of CC on crystallization from solution depends on the equality of the solubility of API and CCF. The crystallization experiments using various solvents are needed to find the suitable solvent, which leads the equality of the solubility of API and CCF. The CCG method does not require the solvent selection because the ability of the CC crystallization complies with the interaction between API and CCFs. However, if the API is too rigid to be changed the crystallinity by grinding, it is difficult to adapt the CCG method. The CC screening by crystallization from solution should be performed for such rigid API. 3.7. Scalability of CCG All the CCF mixtures used in CCG, except for Others, include four CCFs that have the same partial structure. The method enables us to identify the partial structure of the CCFs that preferentially forms synthons with an API. If a CC formed by the CCG method has undesirable physicochemical properties, we can extend the screening study to other CCFs having the same partial structure as the undesirable CC and use the same synthon with the API to find desirable CCs. Thus, CCG is a synthon-oriented method for screening CCs.
4. Conclusion We successfully developed a CCG method, in which an API is ground with CCF mixtures, with an aim of investigating the possibility of CC formation of an API. CCG is an efficient, timesaving, simple, and reliable grinding method, as opposed to the conventional single grinding method, because grinding in the former case enables the formation of CCs and synthons. Since CCG is a synthon-oriented screening method, the CC screening can be further extended to a large number of CCs on the basis of the synthons found in this CCG method. The CCG method enables to find the CC-forming ability of APIs efficiently. This makes the possibility improving the developability of a drug candidate extremely high. Acknowledgements The authors gratefully acknowledge Dr. Takashi Kojima, Drug metabolism and Pharmacokinetics Research Laboratories, Takeda Pharmaceutical Co., Ltd., for his insight and dedicated support. We also like to thank Dr. Toshiya Moriwaki and members of Physicochemistry at DMPK Research laboratories of Takeda Pharmaceutical Co., Ltd., for kind assistance and discussion. References Aakeröy, C.B., Beatty, A.M., Helfrich, B.A., 2001. Total synthesis supramolecular style: design and hydrogen-bond-directed assembly of ternary supermolecules. Angew. Chem. Int. Ed. 40, 3240–3242. Ainouza, A., Authelina, J-R., Billot, P., Liebermanb, H., 2009. Modeling and prediction of cocrystal phase diagrams. Int. J. Pharm. 374, 82–89. Almarsson, Ö., Zaworotko, M.J., 2004. Crystal engineering of the composition of pharmaceutical phases. Do pharmaceutical co-crystals represent a new path to improved medicines? Chem. Commun. 17, 1889–1896. Basavoju, S., Bostrom, D., Velaga, S.P., 2008. Indomethacin–saccharin cocrystal: design, synthesis and preliminary pharmaceutical characterization. Pharm. Res. 25, 530–541. Chiarella, R.A., Davey, R.J., Peterson, M.L., 2007. Making co-crystals the utility of ternary phase diagrams. Cryst. Growth Des. 7, 1223–1226. Chieng, N., Hubert, M., Saville, D., Rades, T., Aaltonen, J., 2009. Cocrystals prepared by mechanical activation formation kinetics and stability of carbamazepine–nicotinamide. Cryst. Growth Des. 9, 2377–2386. Childs, S.L., Hardcastle, K.I., 2007. Cocrystals of piroxicam with carboxylic acids. Cryst. Growth Des. 7, 1291–1304. Childs, S.L., Rodriguez-Hornedo, N., Reddy, L.S., Jayasankar, A., Maheshwari, C., McCausland, L., Shipplett, R., Ctahly, B.C., 2008. Screening strategy based on solubility and solution composition generate pharmaceutically acceptable cocrystals of carbamazepine. Cryst. Eng. Commun. 10, 856–864. ´ T., Jones, W., 2009. Recent advances in understanding the mechanism of Friˇscˇ ic, cocrystal formation via grinding. Cryst. Growth Des. 9, 1621–1637. Fukami, T., Furuishi, T., Suzuki, T., Hidaka, S., Ueda, H., Tomono, K., 2006. Improvement in solubility of poorly water soluble drug by cogrinding with highly branched cyclic dextrin. J. Inclusion Phenom. Macrocycl. Chem. 56, 61–64.
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171 Hickey, M.B., Peterson, M.L., Scoppettuolo, L.A., Morrisette, S.L., Vetter, A., Hector Guzman´ Remenar, J.F., Zhang, Z., Tawa, M.D., Haley, S., Zaworotko, M.J., Almarsson, Ö., 2007. Performance comparison of a co-crystal of carbamazepine with marketed product. Eur. J. Pharm. Biopharm. 67, 112–119. Jayasankar, A., Somwangthanaroj, A., Shao, Z.J., Rodriguez-Hornedo, N., 2006. Cocrystal formation during cogrinding and storage is mediated by amorphous phase. Pharm. Res. 23, 2381–2392. ´ T., Jones, W., Motherwell, W.D.S., 2007. Screening for pharmaceuKarki, S., Friˇscˇ ic, tical cocrystal hydrates via neat and liquid-assisted grinding. Mol. Pharm. 4, 347–354. Kojima, T., Tsutsumi, S., Yamamoto, K., Ikeda, Y., Moriwaki, T., 2010. High-throughput cocrystal slurry-screening by use of in situ raman microscopy and multi-well plate. Int. J. Pharm. 399, 52–59. Lee, H-G., Zhang, G.G.Z., Flanagan, D.R., 2011. Cocrystal intrinsic dissolution behavior using a rotating disk. J. Pharm. Sci. 100, 1727–1735. Limwikrant, W., Higashi, K., Yamamoto, K., Moribe, K., 2009. Characterization of ofloxacin–oxalic acid complex by PXRD, NMR and THz spectroscopy. Int. J. Pharm. 382, 50–55. McNamara, D.P., Childs, S.L., Giordano, J., Iarriccio, A., Cassidy, J., Shet, M.S., Mannion, R., O’Donnell, E., Park, A., 2006. Use of a glutaric acid cocrystal to improve oral bioavailability of a low solubility API. Pharm. Res. 23, 1888–1897. Remenar, J.F., Morissette, S.L., Peterson, M.L., Moulton, B., MacPhee, J.M., Héctor, R., Guzmán, H.R., Almarsson, Ö., 2003. Crystal engineering of novel cocrystals of a triazole drug with 1,4-dicarboxylic acids. J. Am. Chem. Soc. 125, 8456–8457.
171
´ Rodriguez-Hornedo, N., Nehm, S.J., Seefeldt, K.F., Pagan-Torres, Y., Falkiewicz, C.J., 2005. Reaction crystallization of pharmaceutical molecular complexes. Mol. Pharm. 3, 362–367. Schartman, R.R., 2009. On the thermodynamics of cocrystal formation. Int. J. Pharm. 365, 77–80. Schultheiss, N., Newman, A., 2009. Pharmaceutical cocrystals and their physicochemical properties. Cryst. Growth Des. 9, 2950–2967. ˇ F., Rodriguez-Hornedo, N., 2007. CrystallizaSeefeldt, K., Miller, J., Alvarez-Núnez, tion pathways and kinetics of carbamazepine–nicotinamide cocrystals from the amorphous state by in situ thermomicroscopy, spectroscopy and calorimetry studies. J. Pharm. Sci. 96, 1147–1158. Shan, N., Zaworotko, M.J., 2008. The role of cocrystals in pharmaceutical science. Drug Discov. Today 13, 440–446. Takata, N., Shiraki, K., Takano, R., Hayashi, Y., Terada, K., 2008. Cocrystal screening of stanolone and mestanolone using slurry crytallization. Cryst. Growth Des. 8, 3032–3037. Takata, N., Takano, R., Uekusa, H., Hayashi, Y., Terada, K., 2010. A spironolactonesaccharin 1:1 cocrystal hemihydrate. Crys. Growth Des. 10, 2116–2122. Trask, A.V., Motherwell, W.D.S., Jones, W., 2006. Physical stability enhancement of theophylline via cocrystallization. Int. J. Pharm. 320, 114–123. Tsutsumi, S., Iida, M., Tada, N., Kojima, T., Ikeda, Y., Moriwaki, T., Higashi, K., Moribe, K., Yamamoto, K., 2011. Characterization and evaluation of miconazole salts and cocrystals for improved physicochemical properties. Int. J. Pharm. 421, 230–236. Vishweshwar, P., McMahon, J.A., Bis, J.A., Zaworotko, M.J., 2006. Pharmaceutical cocrystals. J. Pharm. Sci. 95, 499–516.
Contents lists available at SciVerse ScienceDirect
International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm
Establishment of cocrystal cocktail grinding method for rational screening of pharmaceutical cocrystals Katsuhiko Yamamoto a,∗ , Shunichirou Tsutsumi b , Yukihiro Ikeda a a Drug Metabolism and Pharmacokinetics Research Laboratories, Pharmaceutical Research Division, Takeda Pharmaceutical Company Limited, 2-26-1, Muraoka-Higashi, Fujisawa, Kanagawa, 251-8555, Japan b Pharmaceutical Technology Research and Development Laboratories, CMC Center, Takeda Pharmaceutical Company Limited, 2-17-85, Juso-Hommachi, Yodogawa-ku, Osaka, 532-8686, Japan
a r t i c l e
i n f o
Article history: Received 20 March 2012 Received in revised form 2 July 2012 Accepted 22 July 2012 Available online 31 July 2012 Keywords: Cocrystal Grinding Drug substance Amorphous Crystallinity Physicochemistry
a b s t r a c t Cocrystals (CCs) used in the pharmaceutical industry are defined as complex crystals formed by reaction between an API and a cocrystal former (CCF); unlike salts, CCs do not show proton transfer. Recently, pharmaceutical CCs have been used to improve the drug-likeness of APIs, such as solubility and stability. Grinding is more effective for CC synthesis than crystallization from solution because in the former case, the API can predominantly interact with the CCF without being affected by solvents. However, this method is tedious because the API is ground with only one CCF at a time. We developed a cocktail cocrystal grinding (CCG) method, in which a mixture of CCFs having the same functional group was used. No false negatives/positives were observed in CCG when carbamazepine was used as the model compound. This method could be used to obtain CCs of piroxicam and spironolactone. False negatives were observed for only one compound from among three model compounds, indicating that CCG facilitates efficient CC detection and that it has higher throughput than does the conventional method. Further, CCG is fast and suitable for rational CC screening, and it helps identify the partial structure of CCFs that forms synthons with an API. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Recently, cocrystals (CCs) of drug substances have been studied extensively with the aim of improving the physicochemical properties of drug substances: solubility, stability, hygroscopicity, etc. (Vishweshwar et al., 2006; Shan and Zaworotko, 2008; Schultheiss and Newman, 2009; Lee et al., 2011). A CC is defined as a complex crystal whose lattice includes more than two chemicals that are in the solid state at room temperature (Almarsson and Zaworotko, 2004). Although salts can be classified as CCs, they show proton transfer, while the CCs fabricated in this study do not. In a CC lattice, a host molecule (drug) interacts with guest molecules, cocrystal formers (CCF), via hydrogen bonding and other weak interactions (Aakeröy et al., 2001; Remenar et al., 2003; Limwikrant et al., 2009). Various methods for crystallizing CCs have been investigated, such as crystallization from solution (Ainouza et al., 2009;
Abbreviations: CC, cocrystal; CCF, cocrystal former; API, active pharmaceutical ingredient; PXRD, powder X-ray diffraction; TG, thermogravimetry; DSC, differential scanning calorimetry. ∗ Corresponding author. Tel.: +81 466 32 2850; fax: +81 466 29 4432. E-mail addresses: Yamamoto [email protected], [email protected] (K. Yamamoto). 0378-5173/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ijpharm.2012.07.038
Rodriguez-Hornedo et al., 2005; Basavoju et al., 2008), crystallization from slurry (Hickey et al., 2007; Takata et al., 2008; Kojima et al., 2010), and melt recrystallization (Seefeldt et al., 2007). Among these, the most popular method is crystallization from solution, which includes cooling, evaporation, and addition of an antisolvent. However, it is often difficult to crystallize CCs from solution because of the following reasons: (1) the large difference between the solubilities of the drug and the CCF (Chiarella et al., 2007); (2) the strong interaction with solvents; (3) and nonattainment of the critical activity to form CCs (Schartman, 2009). Grinding is one of the important methods used to crystallize CCs (Trask et al., 2006; Fukami et al., 2006; Karki et al., 2007; Jayasankar et al., 2006; Friˇscˇ ic´ and Jones, 2009). A schematic illustration of crystallization by grinding is shown in Fig. 1. This method has advantages over crystallization from solution because in the former case, interactions between the drug and the CCFs occur in the solid state. Some of these advantages are as follows: (1) the interaction between the solvent and the drug (or CCF) does not disturb the interaction between the drug and the CCF; (2) since there is no solubility limit in the system, the CCF can be added to the system until the activity of the drug is sufficient for interaction with the CCF and formation of CCs with the API. However, grinding is time-consuming and tedious because each of the CCFs being used needs to be ground with the drug individually. In this study, we
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
163
2.3. Powder X-ray diffraction (PXRD) About 1 mg of the powder was placed on a slide glass and analyzed with a powder X-ray diffractometer, Bruker D8 Discover HTS with GADDS (Madison, WI, USA). The instrument parameters were as follows: X-ray source power, 40 kV/40 mA; collimator radius, 0.3 mm; detector distance, 25 cm; exposure time, 120 s; and detection angle range (2), 5–40◦ .
Fig. 1. Schematic of CC formation by grinding.
2.4. Thermal analysis (TG/DSC) established cocktail cocrystal grinding (CCG), which is a rational method for CC screening. This novel method enables four CCFs to be ground with API simultaneously and hence is more advantageous than the conventional method. The CCFs in the mixture have the identical chemical moieties that interact with the chemical moiety of the drug molecule and form synthons (the minimum unit of interaction between a drug and a CCF). Therefore, the CCG method decreases time, labor, and the amount of API that is to be ground with multiple CCFs simultaneously; further, it enables us to identify the partial structure of the CCFs that preferentially forms synthon with the drug. 2. Materials and methods 2.1. Materials Carbamazepine, piroxicam was obtained from Wako Pure Chemical Industries (Osaka, Japan). Spironolactone, lactamide, ethylmaltol and N-methyl-d-glucamine were obtained from Tokyo Kasei Kogyo (Tokyo, Japan). Glycoamide were obtained from Sigma–Aldrich (St. Louis, MO, USA). The other cocrystal formers were obtained from Wako Pure Chemical Industries (Osaka, Japan). All solvents were purchased from Wako Pure Chemical Industries (Osaka, Japan). 2.2. Grinding operation About 20 mg of powder containing an API and CCFs and a 7-mm stainless-steel ball were placed in a 1.5-mL stainless-steel cell from Retsch (Haan, Germany). Then, 20 L of n-heptane was added to the cell. The cell was shaken in a Retsch MM301 mixer mill (Haan, Germany) at 28.5 rps for 30 min.
About 2 mg of the powder was sampled and placed in an open aluminum crucible and analyzed with a Mettler-Toredo TGA/DSC1 simultaneous TG/DSC analyzer (Greifensee, Switzerland). The instrument parameters were as follows: ramp rate, 5 ◦ C/min; temperature range, 25–300 ◦ C; atmosphere, nitrogen gas; and gas flow rate, 40 mL/min. 3. Results and discussion 3.1. Design and properties of CCF mixture Five CCF mixtures were designed, as shown in Table 1. Each CCF mixture, except the Others group, contains four equimolar CCFs having the same chemical moiety to form synthons. The five CCF mixtures were prepared with various specifications. First, the safety and feasibility of using these CCFs was considered, because these CCFs are excipients, additives, or counter-ions of pharmaceutical salts used in commercially available drug substances and products. Second, these CCFs have been referred to in publications on CCs. CCs synthesized from organosulfonic acids have often been reported, however organosulfonic acids have not been adopted in the present study because they are highly deliquescent and the alcoholic esters of these acids may be genotoxic. Third, CCFs with high aqueous solubility were selected. Some CCs obtained from an API and such CCFs improved the aqueous solubility of the API itself (McNamara et al., 2006; Tsutsumi et al., 2011). During the grinding process, two CCFs in the mixture can react to form CCs. Hence, the CCFs were gently micronized and mixed in a mortar and pestle to avoid mechanochemical changes in the crystallinity and the crystal form. Grinding of the CCF mixtures was carried out solely to identify whether CCs can be formed from
Table 1 Physicochemical properties of CCF mixtures and individual CCFs. CCFs Mixture
CCF Name
Molecular weight (Da)
H-Acceptor
H-Donor
Melting point (◦ C)
Acid 1 (Di-, tri-carboxylic acid)
Citric acid Fumaric acid Succinic acid l-Tartaric acid
192 116 118 151
7 4 4 6
4 2 2 4
135 210 (sublimation) 185 168
Acid 2 (Aromatic, large carboxylic acid)
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
122 138 154 132
2 3 4 4
1 2 3 2
121 159 199 (degradation) 95
Amide (amide)
Nicotinamide Benzamide Glycolamide Lactamide
122 121 75 89
3 2 3 3
1 1 2 2
129 128 122 120
Amine (amine)
Tromethamine Megluimine l-Arginine l-lysine
121 195 174 146
4 6 6 4
4 6 5 3
167 129 222 215
Others (other CCFs)
Urea Ethylmaltol Saccharin Stearic acid
60 140 183 284
3 3 4 2
2 1 1 1
132 (sublimation) 85 (sublimation) 226 (sublimation) 67 (sublimation)
164
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 2. PXRD patterns of CCF mixtures before and after grinding. Inset table shows a detailed annotation of each pattern.
the CCF mixture or not. The PXRD patterns of the CCF mixtures before and after grinding are shown in Figs. 2a (Acid 1 and Acid 2), Fig. 2b (Amide and Amine), and Fig. 2c (Others). No new diffraction peaks were identified in the PXRD patterns of Acid 2 and Amide after grinding. On the other hand, minor diffraction peaks appeared in the PXRD patterns of Acid 1, Amine, and Others after grinding. These new peaks might be attributed to CCs formed by the reaction between CCFs, but the amount of CCs formed was very less. Hence, contamination by these CCs would not affect the formation of CCs between the API and the CCFs, and the resulting CCs could be detected by PXRD. 3.2. Trial CCG with model API (1) – carbamazepine (CBZ) chose carbamazepine (5H-dibenz[b,f]azepine-NWe carboxamide; Fig. 3) as one of the model compounds for CCG. Carbamazepine (CBZ) is a well known and widely used anticonvulsant marketed under the name Tegretol® . Since CBZ is a
Fig. 3. Chemical structure of CBZ.
neutral compound with low solubility in aqueous media, several studies have been carried out to improve its solubility. Homoor heterosynthons can be formed from the amide group of CBZ and the partial structures of the CCFs in CCs (Schartman, 2009). Since the CCs of CBZ were extensively investigated using various crystallization methods (Childs et al., 2008), CBZ is suitable for the model compound of CC. The PXRD patterns for CBZ–CCF mixtures after grinding are depicted in Fig. 4a (CBZ-Acid 1 and CBZ-Acid 2), Fig. 4b (CBZ-Amide and CBZ-Amine), and Fig. 4c (CBZ-Others). Several new diffraction peaks appeared in the PXRD patterns of CBZ-Acid 1, CBZ-Acid 2, CBZ-Amide, and CBZ-Others after grinding. No new diffraction peaks were identified in the PXRD pattern of CBZ-Amine. This indicated CC formation between CBZ and the CCFs in Acid 1, Acid 2, Amide, and Others. We ground CBZ with individual CCFs in the CCF mixture and closely scrutinized the grinding process, to confirm CC formation and identify the CCF that formed CCs with CBZ. If some of the CCFs in the CCF mixture form CCs with CBZ, the mixture would also form the same CCs with CBZ. Acid 1, Acid 2, Amide, and Others were thought to contain CCFs that could form CCs with CBZ. Fig. 5a–d show the PXRD patterns of CBZ with the individual CCFs in Acid 1, Acid 2, Amide, and Others after grinding, respectively. In the case of Acid 1 group, a few unique peaks were identified in the patterns of CBZ-fumaric acid, CBZ-succinic acid, and CBZ-ltartaric acid. These data indicated that CBZ forms CCs with fumaric acid, succinic acid and l-tartaric acid. Similarly, in the case of the Acid 2 group, a few unique peaks were identified in the PXRD patterns of CBZ-benzoic acid, CBZ-salicylic acid, CBZ-gentisic acid, and CBZ-l-glutaric acid. These data indicated that CBZ also formed CCs
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
165
Fig. 4. PXRD patterns of CBZ–CCF mixtures after grinding. Inset table shows a detailed annotation of each pattern.
with benzoic acid, salicylic acid, gentisic acid, and l-glutaric acid. The carboxamide group of CBZ easily forms a heterosynthon with a carboxylic acid that links an aromatic ring. Grinding CBZ with each CCF in Amide resulted in the formation of CBZ-nicotinamide and CBZ-benzamide CCs. Aromatic amides
preferentially form synthons with the carboxamide of CBZ, but aliphatic amides do not form CCs with CBZ. Grinding CBZ with individual CCFs in Others resulted in the formation of CBZ-urea and CBZ-saccharin CCs. CBZ-urea is a novel CC of CBZ.
Table 2 Results summary of CCG method using three model compounds, CBZ, PRX and SPI. Carbamazepine
Piroxicam
Spironolactone
CCF mixture
Individual CCF
CCF mixture
Individual CCF
CCF mixture
Individual CCF
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid1
Citric acid Fumaric acid Succinic acid l-Tartaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Acid2
Benzoic acid Salicylic acid Gentisic acid Glutaric acid
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amide
Nicotinamide Benzamide Glycolamide Lactamide
Amine
Tromethamine Meglumine l-Arginine l-lysine
Amine
Tromethamine Meglumine l-Arginine l-lysine
Amine
Tromethamine Meglumine l-Arginine l-lysine
Others
Urea Ethylmaltol Saccarin Stearic acid
Others
Urea Ethylmaltol Saccarin Stearic acid
Others
Urea Ethylmaltol Saccarin Stearic acid
Bold: hit and reported CC, bold and underline: novel CC, normal: no CC.
166
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 5. PXRD patterns of CBZ-individual CCFs in CCF mixtures after grinding. Inset table shows a detailed annotation of each pattern.
To confirm if any false negatives are observed in this method, we carried out grinding of each CCF in Amine with CBZ. Fig. 5e shows the PXRD patterns of the individual CCFs in Amine: tromethamine, meglumine, l-lysine, and l-arginine. No new peak was identified in the patterns, indicating that CBZ did not form CCs with these amines. From these observations, it was clear that the carboxamide of CBZ does not form heterosynthons with amines. The CCG method does not give false negatives, hence, new CCs of APIs can be identified accurately. In addition, there were no false positives in the trial performed using CBZ. 3.3. Trial CCG with model APIs (2) – piroxicam (PRX) and spironolactone (SPI) To verify the adaptability of CCG method, we tested a nonsteroidal anti-inflammatory drug, piroxicam (PRX, 4-hydroxy2-methyl-N-pyridin-2-yl-2H-1,2-benzothiazine-3-carboxamide 1,1-dioxide) and spironolactone (SPI, 7␣-acetylthio-3-oxo-17␣pregn-4-ene-21,17-carbolactone), which is a diuretic. The chemical
structures of PRX and SPI are shown in Fig. 6. PRX has a carbonyl group, which can form synthons with the partial structures of CCFs. Several CCs of piroxicam have been identified in a previous study (Childs and Hardcastle, 2007). SPI is a neutral compound, and it has some carbonyl groups that form synthons with CCF. Since the aqueous solubility of SPI is low, the improvement of the solubility is required to increase the bioavailability of this API. We assumed that CCs of PRX and SPI were detected by CCG, as CCs of CBZ were clearly identified by this method. Table 2 summarizes the results of CCG trials using three model compounds (CBZ, PRX and SPI). On the PRX trial, new diffraction peaks did not appear in the PXRD patterns of PRX-Acid 1, PRX-Amide, and PRX-Amine, while some new diffraction peaks were identified in the PXRD patterns of PRX-Acid 2 and PRX-Others. These observations confirmed that PRX formed CCs with some of the CCFs in Acid 2 and Others. To confirm the CC formation between PRX and the CCFs and evaluate false positives/negatives in the proposed method, we ground PRX with individual CCFs. Some new diffraction peaks were observed in the PXRD patterns of PRX-citric acid and PRX-succinic acid (from
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
167
Fig. 6. Chemical structure of PRX and SPI.
Acid 1); PRX-benzoic acid, PRX-salicylic acid, PRX-gentisic acid, and PRX-glutaric acid (from Acid 2); and PRX-ethyl maltol (from Others). These results suggested that PRX formed CCs with these CCFs. The formation of PRX-citric acid and PRX-succinic acid CCs was considered as false negatives because PRX did not form CCs with the Acid 1 CCF mixture. PRX-ethyl maltol is a novel CC of PRX. The representative PXRD patterns of PRX with CCF mixtures, Others, and individual CCFs in Others after grinding are depicted in Fig. 7. Next, we ground SPI with CCF mixtures. New diffraction peaks appeared only in the PXRD patterns of PRX-Acid 2. In the case of grinding with the individual CCFs in the CCF mixtures, new diffraction peaks were also observed in the PXRD patterns of SPIbenzoic acid, SPI-salicylic acid and SPI-gentisic acid, which belong to Acid 2 mixture. Fig. 8 depicts the representative PXRD patterns of SPI with CCF mixtures, Acid 2, and individual CCFs in Acid2 after grinding. Therefore, SPI-benzoic acid CC, SPI-salicylic acid CC and SPI-gentisic acid CC were identified. These CCs are novel CCs of SPI. Aromatic carboxylic acids preferentially form synthons with SPI. No false positive/negative was identified. Although SPI-saccharin CCs were reported (Takata et al., 2010), these CCs were not identified in our experiments. In that study, SPI-saccharin CCs was primarily crystallized as hydrates. Since grinding was performed under anhydrous conditions in this study, these CCs were not crystallized.
3.4. Scaleup and characterization of identified CCs We conducted scaleup and characterization of the identified CCs of CBZ, PRX and SPI in order to validate the CC formation and to evaluate physicochemical properties of the CCs. CBZ-urea CC, PRXethyl maltol CC, and SPI-gentisic acid CC were selected from among the CCs formed by CCG. Scaleup was conducted by simply increasing the cell size (1.5–5.0 mL) and sample amount (20–100 mg). The PXRD patterns of the CCs during grinding are shown in Fig. 9a (CBZurea CC), Fig. 9b (PRX-ethyl maltol CC), and Fig. 9c (SPI-gentisic acid CC). The diffraction peaks due to the starting material remained in the PXRD patterns of the ground samples even after 30 min, however a very small amount of the starting materials remained in the ground samples after 60 min. The observations suggested that a longer grinding time was necessary for the scaleup, since crystallization by grinding needs sufficient mechanical force for reaction, amorphization, and mixing (Chieng et al., 2009). The TG/DSC curves of the CCs are depicted in Fig. 10. The melting points of the CBZ-urea and SPI-gentisic acid CCs were different from those of the APIs and CCFs. PRX-ethyl maltol CCs showed a large weight loss with an endothermic event in the temperature range 100–120 ◦ C. The weight loss (about 25%) nearly corresponded to the theoretical amount of ethyl maltol when the PRX/ethyl maltol stoichiometry in PRX-ethyl maltol CC was 1:1. The weight loss was thought to
Fig. 7. PXRD patterns of PRX after grinding. (a) PRX-Others mixture (b) PRX-individual CCFs in Others mixture.
168
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 8. PXRD patterns of SPI after grinding. (a) SPI-Acid 2 mixture (b) SPI-individual CCFs in Acid 2 mixture.
be due to the dissociation of ethyl maltol from the CCs, similar to dehydration of hydrate crystals on heating. After dissociation, the PRX-ethyl maltol CCs were converted to PRX; this conversion was confirmed by the endothermic peak appearing at 201 ◦ C, which is the melting point of PRX form I. 3.5. Effects of dropped solvents into grinding N-Heptane was used as the dropped solvent on the grinding procedure in the CCG method. The effect and role of the dropped solvent was properly examined using CBZ-urea CC, which was
detected in the CCG method and crystallized again in the scaleup. We operated solvent drop grinding of CBZ and urea with the same amount of n-heptane, water and acetone, and without solvents, neat grinding. Water and acetone are good solvent for urea and CBZ, respectively. After 30 min grinding, CBZ-urea CC was detected in all the samples, however, CBZ form III remained in the samples of water, acetone and neat grinding. The residual CBZ was not observed in the sample of n-heptane. When the neat grinding was continued, only CBZ-urea CC was identified in the ground sample after 60 min. It is thought that solvents accelerate the reaction by raising the homogeneity and flowability of the powder mixture;
Fig. 9. PXRD patterns of novel CCs on scaleup experiments. Inset table shows a detailed annotation of each pattern.
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
169
Fig. 10. TG (a, c, e) and DSC (b, d, f) curves of APIs, CCFs and novel CCs. (a and b) CBZ; (c and d) PRX, (e and f) SPI.
however, good solvents disturb the reaction between API and CCF due to the dissolving ability, i.e. interaction with API or CCF. Therefore, n-heptane, which is a poor solvent for most of APIs and CCFs, is appropriate to grinding CC crystallization. Especially, because the CCG method uses the CCF mixtures that include various CCFs having each diverse solubility, n-heptane is very beneficial as the dropped solvent. 3.6. Rationality of CCG The rationality of CC screening by CCG should be discussed in comparison with the conventional single (one-to-one) grinding method. In the case of CC screening with the conventional method, 20 grinding experiments were required to identify one CC from among the CCs formed between the API and the 20 selected CCFs. On the other hand, when using CCG, only nine experiments were required to identify one CC; these experiments included five experiments with CCF mixtures and four experiments with the individual
CCFs in the mixtures. A comparison of the conventional method with CCG is illustrated in Fig. 11. The workload for CC screening in CCG is about 50% of that in the single method. Additionally, in CCG, only small amounts of APIs are required. The sample amount of API is often limited in early stage of preclinical study. The CCG method is relevant to CC screening in drug discovery. The reliability of the CCG method against the single method should also be considered. From among the three model compounds, false negatives were identified in only two cases: PRX-citric acid CCs and PRX-succinic acid CCs. On the contrary, no false positive was observed. The following issues need to be addressed in the case of CCG: (1) CC formation between a CCF and the API may compete with CC formation between some CCFs; (2) ternary CCs such as API-CCF 1CCF 2 may be formed by the interaction of two different CCs formed from the API and a CCF. The former competition is thought to block CC formation between the API and a single CCF, which may lead to a false negative. Examples of this blocking could be the formation of
170
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171
Fig. 11. Comparison of conventional method and CCG method.
PRX-citric acid CC and PRX-succinic acid CC in the trial performed on PRX. The latter ternary complex may produce a false positive. In the CCG method, when CC formation is identified on grinding an API with the CCF mixture, CCs are invariably identified on grinding with the individual CCFs. False positives were not observed in this study. Furthermore, the pros and cons of the CCG method are considered. Comparing with single grinding, multiple CCFs compete against the interaction with API molecule in the CCG method. The CCG method is easy to detect CCs that have physically stronger interaction with API. The strength of the interaction was not known in single grinding method. However, the CCG method might miss the CCs that have the weak interaction. Such CCs are able to be favorably detected by single grinding method. In fact, PRXcitric acid CC and PRX-succinic acid CC were not identified in the CCG method trials. From the point of the structure of CC, single grinding might presume the stoichiometry of a API and a CCF in a CC. On the other hand, the CCG method cannot anticipate it because the detected CCs in the CCG method are the mixture of CCs. Next, the comparison with crystallization from solution is discussed. If the CC screening by crystallization from solution using CCF mixture is performed, the detected CCs are thought to be easily crystallized CCs, not the CCs that have the physically strong interaction as detected in the CCG method. The crystallization ability of CC on crystallization from solution depends on the equality of the solubility of API and CCF. The crystallization experiments using various solvents are needed to find the suitable solvent, which leads the equality of the solubility of API and CCF. The CCG method does not require the solvent selection because the ability of the CC crystallization complies with the interaction between API and CCFs. However, if the API is too rigid to be changed the crystallinity by grinding, it is difficult to adapt the CCG method. The CC screening by crystallization from solution should be performed for such rigid API. 3.7. Scalability of CCG All the CCF mixtures used in CCG, except for Others, include four CCFs that have the same partial structure. The method enables us to identify the partial structure of the CCFs that preferentially forms synthons with an API. If a CC formed by the CCG method has undesirable physicochemical properties, we can extend the screening study to other CCFs having the same partial structure as the undesirable CC and use the same synthon with the API to find desirable CCs. Thus, CCG is a synthon-oriented method for screening CCs.
4. Conclusion We successfully developed a CCG method, in which an API is ground with CCF mixtures, with an aim of investigating the possibility of CC formation of an API. CCG is an efficient, timesaving, simple, and reliable grinding method, as opposed to the conventional single grinding method, because grinding in the former case enables the formation of CCs and synthons. Since CCG is a synthon-oriented screening method, the CC screening can be further extended to a large number of CCs on the basis of the synthons found in this CCG method. The CCG method enables to find the CC-forming ability of APIs efficiently. This makes the possibility improving the developability of a drug candidate extremely high. Acknowledgements The authors gratefully acknowledge Dr. Takashi Kojima, Drug metabolism and Pharmacokinetics Research Laboratories, Takeda Pharmaceutical Co., Ltd., for his insight and dedicated support. We also like to thank Dr. Toshiya Moriwaki and members of Physicochemistry at DMPK Research laboratories of Takeda Pharmaceutical Co., Ltd., for kind assistance and discussion. References Aakeröy, C.B., Beatty, A.M., Helfrich, B.A., 2001. Total synthesis supramolecular style: design and hydrogen-bond-directed assembly of ternary supermolecules. Angew. Chem. Int. Ed. 40, 3240–3242. Ainouza, A., Authelina, J-R., Billot, P., Liebermanb, H., 2009. Modeling and prediction of cocrystal phase diagrams. Int. J. Pharm. 374, 82–89. Almarsson, Ö., Zaworotko, M.J., 2004. Crystal engineering of the composition of pharmaceutical phases. Do pharmaceutical co-crystals represent a new path to improved medicines? Chem. Commun. 17, 1889–1896. Basavoju, S., Bostrom, D., Velaga, S.P., 2008. Indomethacin–saccharin cocrystal: design, synthesis and preliminary pharmaceutical characterization. Pharm. Res. 25, 530–541. Chiarella, R.A., Davey, R.J., Peterson, M.L., 2007. Making co-crystals the utility of ternary phase diagrams. Cryst. Growth Des. 7, 1223–1226. Chieng, N., Hubert, M., Saville, D., Rades, T., Aaltonen, J., 2009. Cocrystals prepared by mechanical activation formation kinetics and stability of carbamazepine–nicotinamide. Cryst. Growth Des. 9, 2377–2386. Childs, S.L., Hardcastle, K.I., 2007. Cocrystals of piroxicam with carboxylic acids. Cryst. Growth Des. 7, 1291–1304. Childs, S.L., Rodriguez-Hornedo, N., Reddy, L.S., Jayasankar, A., Maheshwari, C., McCausland, L., Shipplett, R., Ctahly, B.C., 2008. Screening strategy based on solubility and solution composition generate pharmaceutically acceptable cocrystals of carbamazepine. Cryst. Eng. Commun. 10, 856–864. ´ T., Jones, W., 2009. Recent advances in understanding the mechanism of Friˇscˇ ic, cocrystal formation via grinding. Cryst. Growth Des. 9, 1621–1637. Fukami, T., Furuishi, T., Suzuki, T., Hidaka, S., Ueda, H., Tomono, K., 2006. Improvement in solubility of poorly water soluble drug by cogrinding with highly branched cyclic dextrin. J. Inclusion Phenom. Macrocycl. Chem. 56, 61–64.
K. Yamamoto et al. / International Journal of Pharmaceutics 437 (2012) 162–171 Hickey, M.B., Peterson, M.L., Scoppettuolo, L.A., Morrisette, S.L., Vetter, A., Hector Guzman´ Remenar, J.F., Zhang, Z., Tawa, M.D., Haley, S., Zaworotko, M.J., Almarsson, Ö., 2007. Performance comparison of a co-crystal of carbamazepine with marketed product. Eur. J. Pharm. Biopharm. 67, 112–119. Jayasankar, A., Somwangthanaroj, A., Shao, Z.J., Rodriguez-Hornedo, N., 2006. Cocrystal formation during cogrinding and storage is mediated by amorphous phase. Pharm. Res. 23, 2381–2392. ´ T., Jones, W., Motherwell, W.D.S., 2007. Screening for pharmaceuKarki, S., Friˇscˇ ic, tical cocrystal hydrates via neat and liquid-assisted grinding. Mol. Pharm. 4, 347–354. Kojima, T., Tsutsumi, S., Yamamoto, K., Ikeda, Y., Moriwaki, T., 2010. High-throughput cocrystal slurry-screening by use of in situ raman microscopy and multi-well plate. Int. J. Pharm. 399, 52–59. Lee, H-G., Zhang, G.G.Z., Flanagan, D.R., 2011. Cocrystal intrinsic dissolution behavior using a rotating disk. J. Pharm. Sci. 100, 1727–1735. Limwikrant, W., Higashi, K., Yamamoto, K., Moribe, K., 2009. Characterization of ofloxacin–oxalic acid complex by PXRD, NMR and THz spectroscopy. Int. J. Pharm. 382, 50–55. McNamara, D.P., Childs, S.L., Giordano, J., Iarriccio, A., Cassidy, J., Shet, M.S., Mannion, R., O’Donnell, E., Park, A., 2006. Use of a glutaric acid cocrystal to improve oral bioavailability of a low solubility API. Pharm. Res. 23, 1888–1897. Remenar, J.F., Morissette, S.L., Peterson, M.L., Moulton, B., MacPhee, J.M., Héctor, R., Guzmán, H.R., Almarsson, Ö., 2003. Crystal engineering of novel cocrystals of a triazole drug with 1,4-dicarboxylic acids. J. Am. Chem. Soc. 125, 8456–8457.
171
´ Rodriguez-Hornedo, N., Nehm, S.J., Seefeldt, K.F., Pagan-Torres, Y., Falkiewicz, C.J., 2005. Reaction crystallization of pharmaceutical molecular complexes. Mol. Pharm. 3, 362–367. Schartman, R.R., 2009. On the thermodynamics of cocrystal formation. Int. J. Pharm. 365, 77–80. Schultheiss, N., Newman, A., 2009. Pharmaceutical cocrystals and their physicochemical properties. Cryst. Growth Des. 9, 2950–2967. ˇ F., Rodriguez-Hornedo, N., 2007. CrystallizaSeefeldt, K., Miller, J., Alvarez-Núnez, tion pathways and kinetics of carbamazepine–nicotinamide cocrystals from the amorphous state by in situ thermomicroscopy, spectroscopy and calorimetry studies. J. Pharm. Sci. 96, 1147–1158. Shan, N., Zaworotko, M.J., 2008. The role of cocrystals in pharmaceutical science. Drug Discov. Today 13, 440–446. Takata, N., Shiraki, K., Takano, R., Hayashi, Y., Terada, K., 2008. Cocrystal screening of stanolone and mestanolone using slurry crytallization. Cryst. Growth Des. 8, 3032–3037. Takata, N., Takano, R., Uekusa, H., Hayashi, Y., Terada, K., 2010. A spironolactonesaccharin 1:1 cocrystal hemihydrate. Crys. Growth Des. 10, 2116–2122. Trask, A.V., Motherwell, W.D.S., Jones, W., 2006. Physical stability enhancement of theophylline via cocrystallization. Int. J. Pharm. 320, 114–123. Tsutsumi, S., Iida, M., Tada, N., Kojima, T., Ikeda, Y., Moriwaki, T., Higashi, K., Moribe, K., Yamamoto, K., 2011. Characterization and evaluation of miconazole salts and cocrystals for improved physicochemical properties. Int. J. Pharm. 421, 230–236. Vishweshwar, P., McMahon, J.A., Bis, J.A., Zaworotko, M.J., 2006. Pharmaceutical cocrystals. J. Pharm. Sci. 95, 499–516.