Crystengcomm

CrystEngComm

PAPER

Thermodynamics and preliminary pharmaceutical characterization of a melatonin–pimelic acid Cite this: CrystEngComm,2015,17, 612 cocrystal prepared by a melt crystallization method†

Yan Yan,ab Jia-Mei Chen*a and Tong-Bu Luab

Pharmaceutical cocrystals have been extensively investigated as a promising approach to improve drugs' physicochemical properties. Cocrystals are usually prepared using solution-mediated methods and grinding methods. In this study, a cocrystal of pimelic acid with poorly soluble melatonin was produced using a melt crystallization method by controlling the crystallization within a specific temperature range. The cocrystal was characterized by infrared spectroscopy, powder and single crystal X-ray diffraction. The binary phase diagram and the formation enthalpy established by differential scanning calorimetry provided the rationale Received 18th September 2014, for this spontaneous cocrystallization system. The cocrystal exhibited higher apparent solubility, a faster Accepted 1st November 2014 dissolution rate and acceptable stability as compared to the original melatonin. The study has shown that the melt crystallization method can be considered as a competitive route for producing pharmaceutical DOI: 10.1039/c4ce01921k cocrystals with improved properties and the thermodynamic investigation can provide in depth under- www.rsc.org/crystengcomm standing and guidance on the melt crystallization of cocrystals.

the growing number of research publications and patent Introduction – applications.4 12 Pharmaceutical cocrystals can be defined as crystalline Cocrystals are usually prepared using solution- materials composed of an active pharmaceutical ingredient mediated methods (including solvent evaporation, reaction – (API) and one or more pharmaceutically acceptable com- crystallization, slurrying, anti-solvent addition, etc.).13 17 This pounds (guests, coformers), usually held together by reliable approach, however, is limited by the time-consuming solvent – hydrogen bonding interactions, in the same crystal lattice.1 3 selection process, the potential for single component crystal- The molecular packing rearrangements of APIs in cocrystals lization and solvate formation. Solid-based methods (neat or are facilitated by the introduction of guest molecules into the solvent-assisted grinding) have also been demonstrated as crystal lattice, resulting in the modification of their relevant effective methodologies for the preparation of cocrystals.13,18 physicochemical properties. Thus, pharmaceutical cocrystals Grinding methods often need special equipment to keep the provide new opportunities to enhance the physicochemical grinding frequency, and the resulting products often contain properties, such as the melting point, hygroscopicity, mechani- amorphous impurities. Additionally, thermal methods cal properties, permeability, solubility, dissolution rate, and bio- (Kofler's technique, co-melting or melt crystallization), which – availability, of APIs.4 11 In addition, the formation of cocrystals have long been known but rarely used in cocrystal systems, may also create the intellectual property and new patents of are a green alternative approach for identifying cocrystal – APIs, and thus extend their life cycle in the pharmaceutical phases.19 22 This approach could only be applied to thermally industry.12 Hence, pharmaceutical cocrystals have drawn great stable compounds with appropriate melting points. The attention from both academia and industry as evidenced by choice of cocrystallization techniques in cocrystal screening depends on the specific properties of the starting materials, including solubility, crystallinity, melting point, thermal stability, etc. a School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou, China. Melatonin (MT, Fig. 1) ĲN-acetyl-5-methoxytryptamine), a E-mail: [email protected]; Fax: +86 20 84112921; natural product, is found in the pineal gland and excreted as Tel: +86 20 84112921 a hormone in humans.23 It is produced in the brain to help b MOE Key Laboratory of Bioinorganic and Synthetic Chemistry/School of regulate the sleep and wake cycle.24 Exogenous MT is used Chemistry and Chemical Engineering, Sun Yat-sen University, Guangzhou, China † CCDC 1024880. For crystallographic data in CIF or other electronic format see orally for jet lag, insomnia, shift work disorder, circadian DOI: 10.1039/c4ce01921k rhythm disorders in blind individuals, and benzodiazepine

612 | CrystEngComm, 2015, 17,612–620 This journal is © The Royal Society of Chemistry 2015 CrystEngComm Paper

Table 1 The results of melt crystallization screen experiments with different coformers

Coformer MPa/°C Screen resultsb Glycolic acid 80 × Glutaric acid 98 × Oxalic acid 99 × Pimelic acid 103 √ Mandelic acid 119 × Fig. 1 The structures of (a) MT and (b) PA. Benzoic acid 121 × Malic acid 131 × Cinnamic acid 134 × 25 and nicotine withdrawal. In most countries MT prepara- a b √ × 26 Melting point. : formation of a new crystalline phase. : the tions are available as food supplements. A prescription-only physical mixture of MT and each coformer. MT product was approved for use by the European Medicines Agency in 2007.27 MT belongs to Class II of the Bio- pharmaceutics Classification System (BCS) with low solubility Hot-stage microscopy (HSM) 28,29 and high permeability. The oral bioavailability of MT in Thermomicroscopic investigations were performed with an 30 humans is limited and variable. Consequently, improving optical polarizing microscope (Nikon LV100 POL) equipped the solubility of MT is highly desirable. with a Linkam hot stage THMS 600 connected to a THMS Herein we attempted to address the deficiency in solu- 600 temperature controller and an LNP 95 liquid nitrogen bility of MT by cocrystallization with soluble guest mole- pump. The microscopy images were recorded with a CCD cules. Formation of a cocrystal requires complementary camera attached to the Nikon LV100 POL microscope at every hydrogen bonding between MT and a second molecule. The 10 s time interval using the Linksys 32 image capture soft- most useful hydrogen bonding group of MT is the second- ware. A 1 : 1 MT/PA physical mixture was heated to 130 °C, ary amide functionality, which is known to form robust and the temperature was kept until all the solids completely − hydrogen bonding interactions with carboxylic com- melted. Then the melt was cooled to 50 °Catarateof10°Cmin1 31,32 pounds. A group of nontoxic carboxylic acids were used and the temperature was kept at 50 °C until the crystalliza- for cocrystal screening. Finally, only one cocrystal of MT tion process was completed. with pimelic acid (PA, Fig. 1) was successfully obtained through a melt crystallization method and characterized Preparation of single crystals of the MT–PA cocrystal by infrared spectroscopy, powder and single crystal X-ray diffraction. Thermodynamic investigation and pharmaceuti- The products obtained by melt crystallization were used as cal characterization of the MT–PA cocrystal were also raw materials for growing high-quality crystals from solution. – carried out. The powder (20 mg) of the MT PA cocrystal was dissolved in a mixed solvent of dichloromethane (1 mL) and n-hexane (0.6 mL). The resulting solution was allowed to evaporate Materials and methods slowly at room temperature and needle-like single crystals Materials were obtained. MT was purchased from Hubei Yuancheng Pharmaceutical Single crystal X-ray diffraction Co., Ltd. All of the coformers were purchased from Aladdin Reagents Inc. All other chemicals and solvents were commer- Single crystal X-ray diffraction data for the MT–PA cocrystal cially available and used as received. were collected on an Agilent Technologies Gemini A Ultra system with graphite monochromated Cu Kα radiation

Melt crystallization Table 2 The results of melt crystallization experiments at different MT/PA A melt crystallization method was employed to screen a total molar ratios and different holding temperatures after cooling the melt of 8 carboxylic acids and determine whether unique b cocrystalline phases with MT were able to form. In a typical Results a experiment, a physical mixture of MT and each coformer at a Temperature 2 : 1 MT/PA 1 : 1 MT/PA 1 : 2 MT/PA − chosen ratio was heated at 10 °C min 1 until a melt was RT Mixture Mixture Mixture formed. The molten mixture was then cooled to different spe- 30 °C Mixture Mixture Mixture − ° cific temperatures at a cooling rate of 10 °C min 1. Then the 40 C Mixture Mixture Mixture 50 °C Cocrystal + MT Cocrystal Cocrystal + PA mixture was kept at each temperature until the crystallization 60 °C Cocrystal + MT Cocrystal Cocrystal + PA process was completed, which usually takes 10 to 24 h. The 70 °C Cocrystal + MT Cocrystal Cocrystal + PA resulting solids were analyzed by X-ray diffraction for new 80 °C Molten Molten Molten crystalline phases. The results of the screen experiments are a Room temperature (~25 °C). b Mixture: physical mixture of MT listed in Tables 1 and 2. and PA.

This journal is © The Royal Society of Chemistry 2015 CrystEngComm,2015,17,612–620 | 613 Paper CrystEngComm

(λ = 1.54178 Å). Cell refinement and data reduction were Table 4 Hydrogen bond parameters for the MT–PA cocrystal applied using the program of CrysAlisPro.33 The structures D–H⋯AD⋯A (Å) D–H⋯A(°) were solved by direct methods using the SHELX-97 program34 – ⋯ and refined by the full-matrix least-squares method on F2. All O3 H3 O1 2.575(2) 164(4) O6–H6⋯O5a 2.660(2) 172(3) non-hydrogen atoms were refined with anisotropic displace- N1–H1⋯O3b 3.209(3) 175(2) ment parameters. Hydrogen positions on nitrogen and oxy- N2–H2⋯O4c 3.055 125 gen were located in Fourier difference electron density maps. Symmetry codes:a −x +1,−y, −z +1.b 2x − 1, −y + 1/2, z − 1/2. c 1+x, Hydrogen atoms associated with carbon atoms were refined 0.5 − y, 0.5 + z. in geometrically constrained riding positions. Crystallo- graphic data and details of refinements are listed in Table 3 while hydrogen bonding distances and angles are given Dissolution experiments in Table 4. The concentrations of MT were determined by a Cary 50 UV/vis spectrophotometer at 230 nm, where PA does not Powder X-ray diffraction (PXRD) absorb and, therefore, would not interfere with the determination of the concentration of MT. For experiments involving PXRD patterns were obtained on a Bruker D2 Phaser with Cu MT, the MT–PA cocrystal and the 1 : 1 MT/PA physical mix- α λ K radiation ( = 1.54178 Å) at 30 kV and 10 mA. Data were ture, the absorbance values were related to the concentra- – ° θ recorded in the range of 5 40 (2 ) in a continuous scan tions of MT using a calibration curve. ° θ mode using a step size of 0.014 (2 ) and a scan speed of For the powder dissolution experiments, all of the solids 0.1 s per step. were milled to powder and sieved using standard mesh sieves to provide samples with approximate particle size ranges of – μ Fourier transform infrared spectroscopy (FTIR) 75 150 m. In a typical experiment, a flask containing 300 mg of the sample was added to 50 mL of phosphate buff- −1 FTIR spectra were recorded in the 4000 to 400 cm region ered saline (PBS, pH 6.8), and the resulting slurry was stirred using KBr pellets and a Bruker EQUINOX 55 spectrometer. A at 37 °C and 500 rpm. At each time interval an aliquot of the −1 total of 64 scans were collected with a resolution of 0.2 cm slurry was withdrawn from the flask and filtered through a for each sample. 0.22 μm nylon filter. A 0.01 mL portion of the filtered aliquot was diluted to 1.0 mL with water and was measured by Differential scanning calorimetry (DSC) UV/vis spectrophotometry. After the last aliquot was collected, the remaining solids were filtered, dried and analyzed DSC was performed using a Netzsch 200 F3 instrument. by PXRD. – Samples were weighed (1.8 2.5 mg) in aluminum sample Intrinsic dissolution measurements were performed using ° pans and scanned from 30 to 130 C under a nitrogen atmo- the dissolution device ZQY-2 Dissolution Tester (Shanghai ° −1 sphere at a heating rate of 5 C min . Huanghai Yaojian Instrument Distribution Co., Ltd.). In each experiment, 500 mL of PBS (pH 6.8) were equilibrated at 37.0 °C in a constant-temperature bath and stirred at Table 3 Crystallographic data for the MT–PA cocrystal 100 rpm. Approximately 150 mg of each solid was com- MT–PA pressed in a hydraulic press under 1 ton for 5 seconds in a T/K 150(2) die with a 5 mm diameter disk. The disks were coated using Wavelength/Å 1.54184 paraffin wax, leaving only the surface under investigation free Formula C20H28N2O6 for dissolution. These disks were immersed in the dissolu- Formula weight 392.44 tion media and, at each time interval, 2 mL of the solution Crystal system Monoclinic Space group P21/c was withdrawn and replaced by an equal volume of fresh a/Å 5.2497(3) medium to maintain a constant volume. The concentration b/Å 49.977(2) of MT was measured by UV/vis spectrophotometry. Each c/Å 7.6897(5) α/° 90 intrinsic dissolution test lasted 60 min, after which time the β/° 104.573(6) disks were recovered, carefully ground and checked by PXRD. γ/° 90 Finally, the intrinsic dissolution rates were calculated from 3 Ĳ V/Å 1952.58 19) the slopes of the linear regions of the dissolution curves. Z 4 −3 Dcalc/g cm 1.335 Rint 0.0314 Dynamic vapor sorption (DVS) studies GOF 1.123 > σĲ a R1,wR2 [I 2 I)] 0.0498, 0.1160 A DVS intrinsic instrument (Surface Measurement Systems, b R1,wR2 [all data] 0.0623, 0.1225 London, UK) was employed to investigate the hygroscopicity P P P P a b 2 2 2 2 2 1/2 of MT, the MT–PA cocrystal, and the physical mixture by R1 = ∥Fo| − |Fc∥/ |Fo|. wR2 = ĳ ĳwĲFo − Fc ) ]/ wĲFo ) ] , ĳσ2Ĳ 2 2 ĳ 2 2 w=1/ Fo) +(aP) + bP], where P = (Fo )+2Fc ]/3. measuring the uptake and loss of water vapor gravimetrically

614 | CrystEngComm,2015,17,612–620 This journal is © The Royal Society of Chemistry 2015 CrystEngComm Paper using an ultra sensitive recording microbalance with a mass yielding only physical mixtures of individual components. To resolution of ±0.1 μg. The temperature was maintained con- achieve a robust cocrystallization process, it is a prerequisite stant at 25 ± 0.1 °C. For each sample, about 20 mg was to identify key factors affecting the melt crystallization of MT loaded onto a metal sample pan and pre-equilibrated at and PA. Thus a series of melt crystallization experiments at 0% RH in a continuous flow of dry nitrogen for several hours different MT/PA molar ratios and different temperatures to establish the equilibrium dry mass. The relative humidity were conducted (Table 2). The PXRD patterns of the different was then increased from 0 to 90% in 10% steps. For the MT/PA systems are presented in Fig. 2. When a molten mix- desorption process, the RH was decreased in a similar man- ture at a MT/PA molar ratio of 1 : 2 or 2 : 1 was cooled to ner. The equilibration criterion dm/dt (change in mass as a 50 °C and allowed to crystallize completely, new peaks were − function of time) was set at 0.002% min 1 for all steps. The observed other than those corresponding to excess amounts sorption isotherms were calculated from the equilibrium of the starting components (Fig. 2), which indicates that 1 : 2 mass values. and 2 : 1 are unsuitable stoichiometries for cocrystal formation. Meanwhile, no PXRD peaks indicative of the starting Results and discussion components were observed for the crystallization product with the 1 : 1 ratio (Fig. 2), thereby suggesting that the stoi- Preparation of the cocrystal chiometry of the cocrystal is 1 : 1. A set of 8 pharmaceutically acceptable carboxylic acids were To further determine whether the temperature at which selected as potential cocrystal coformers (Table 1). Crystalli- crystallization occurs affects the outcome of melt crystalliza- zation experiments on solutions of MT and guest acids tion, 1 : 1 MT/PA molten mixture was cooled to a series of resulted in individual component phases because of the dif- specific temperatures and then kept at each temperature ferences in their solubility. As the selected carboxylic acids until the crystallization process was completed. We have and MT are thermally stable compounds, we chose to use a tested the temperature from RT (~25 °C) to 80 °C, which is cocrystal screening technique based on a melt crystallization close to the melting point of the MT–PA cocrystal. The method. Most carboxylic acids produced physical mixtures of cocrystal formation between MT and PA apparently occurred the parental compounds when crystallized from the molten when their 1 : 1 molten mixture was allowed to crystallize under mixtures with MT. Only one unique cocrystal of MT with PA a temperature within the range of 50–70 °C. The crystallization (MT–PA cocrystal) was identified by the appearance of new behavior of the 1 : 1 molten mixture of MT/PA at 50 °Cwas reflections in the diffractogram at 7.1, 25.6° (2θ) (Fig. 2) observed by HSM (Fig. 3). Some needle-like crystals were found which do not exist in the diffractograms of the starting to appear from the amorphous phase at 50 °Cafter10min, materials. and the whole crystallization process was completed in 10 h. Development of a robust procedure to prepare the MT–PA The physical mixtures of the components were obtained upon cocrystal was proved to be non-trivial. In initial experiments, crystallization in the 25–40 °C range; on the other hand, it molten mixtures of MT and PA were allowed to cool to room is possible to produce either the cocrystal or the physical temperature (RT) and to crystallize under this temperature, mixture upon crystallization in the 40–50 °C range. Overall, the robust preparation method of the phase pure MT–PA cocrystal involves cooling a 1 : 1 MT/PA molten mixture and controlling the crystallization temperature between 50 and 70 °C.

Crystal structure analysis The crystal structure analysis of the MT–PA cocrystal was performed to identify the molecular interactions between MT – and PA. The MT PA cocrystal adopts the monoclinic P21/c space group, with one molecule each of MT and PA in the asymmetric unit (Table 3). To compare the hydrogen bonding models before and after cocrystal formation, the structures of pure MT and PA were analyzed from a crystal engineering perspective. Examination of the crystal structure of MT (refcode MELATN01) reveals that the secondary amide group is the main hydrogen bond donor–acceptor functionality which links MT molecules to form a one-dimensional chain Fig. 2 Comparison of the PXRD patterns of the MT–PA system: through N–H⋯O hydrogen bonds (N⋯O, 2.963 Å; N–H⋯O, – ° (a) MT PA cocrystal (calculated); (b) 1 : 1 MT/PA crystallized at 50 C; 161.62°) (Fig. 4a). The crystal structure of PA (refcode (c) MT; (d) PA; (e) 1 : 2 MT/PA crystallized at 50 °C; (f) 2 : 1 MT/PA crystallized at 50 °C; and (g) 1 : 1 MT/PA crystallized at RT. In (b) the PIMELA06) adopts a structural unit of carboxylic acid dimers pattern of the bulk powder is in good agreement with the pattern via double O–H⋯O hydrogen bonds (O⋯O, 2.669 Å; calculated from the single-crystal structure (a). O–H⋯O, 175.52°) (Fig. 4b). After the formation of the MT–PA

This journal is © The Royal Society of Chemistry 2015 CrystEngComm,2015,17,612–620 | 615 Paper CrystEngComm

Fig. 3 The crystallization behavior of the 1 : 1 molten mixture of MT/PA at 50 °C after (a) 0 min, (b) 10 min, (c) 12 min, (d) 30 min, (e) 4 h and (f) 10 h.

cocrystal, one carboxylic acid group of PA links to the amide bonded carboxylic acid dimers form in the PA crystalline functionality of MT by a pair of O3–H3⋯O1 and N1–H1⋯O3 lattice. hydrogen bonds (O3⋯O1, 2.575(2) Å; O3–H3⋯O1, 164Ĳ4)°; The spectrum of the MT–PA cocrystal has three distinctive − N1⋯O3, 3.209(3) Å; N1–H1⋯O3, 175Ĳ2)°, Table 4) to form a peaks at 1638, 1691 and 1711 cm 1, which are assigned to one-dimensional chain. This carboxylic acid group also forms CO stretching of the amide group in MT and the two car- a strongly bent and thus weaker hydrogen bond (N2⋯O4, boxylic acid groups in PA, respectively (Fig. 5c). The increase − − 3.055 Å; N2–H2⋯O4, 125°) with the ring N–H donor of MT. in the CO stretching frequency from 1630 cm 1 and 1689 cm 1 − The second acid group retains an acid–acid dimer structural in the component phases, respectively, to 1638 cm 1 and − unit with double O6–H6⋯O5 hydrogen bonds (O⋯O, 1711 cm 1 in the cocrystal implies that the CO groups par- 2.660(2) Å; O–H⋯O, 172Ĳ3)°, Table 4), which make the one- ticipate in weaker hydrogen bonds accompanying cocrystal dimensional chains assemble into a two-dimensional sheet formation. The O–H stretching frequency of PA and the N–H (Fig. 4c). The hydrogen bonding distance of the amide group stretching frequency of MT were observed at 3390 and − increases from 2.963 Å in MT to 3.209(3) Å in the MT–PA 3435 cm 1, respectively, in the cocrystal (Fig. 5c). A cocrystal. The hydrogen bonding angle of the carboxylic acid hypsochromic shift in the O–H and N–H stretching frequency decreases from 175.52° in PA to 164Ĳ4)° in the cocrystal. further verifies the formation of weaker hydrogen bonds These results indicate that as MT cocrystallizes with PA, the accompanying cocrystallization. These results from the bulk resulting structure fails to achieve stronger hydrogen bond- cocrystal are consistent with the observations from single ing interactions and the formed hydrogen bonds are weaker crystal structure analysis. than those in the individual component phases.

Thermodynamics of cocrystal formation FTIR spectroscopy characteristics After characterization of the cocrystal, we attempted to inves- The FTIR spectra of MT, PA and the MT–PA cocrystal are tigate the thermodynamics of cocrystal formation. The tem- presented to further verify the hydrogen bonding changes perature–composition phase diagram is essential knowledge involved in the bulk cocrystal production (Fig. 5). MT shows for any cocrystal system as it indicates the phases formed at − characteristic absorption peaks at 1630 and 3304 cm 1, which different compositions of the cocrystal formers over a range are assigned to CO and N–H stretching of the secondary of temperatures. The phase diagram of the MT–PA system amide, respectively (Fig. 5a). This indicates that MT is the was constructed by thermal analysis of the corresponding starting material in the cocrystallization. The spectrum of PA binary mixtures using DSC (Fig. 6). The present study shows has peaks corresponding to carboxylic CO and O–H a representative phase diagram of a binary mixture forming a − − stretching at 1689 cm 1 and 2939 cm 1, respectively (Fig. 5b). 1 : 1 cocrystal with three local melting point maxima (i.e., for This suggests that the structural units of the hydrogen the two pure components and the cocrystal) separated by two

616 | CrystEngComm,2015,17,612–620 This journal is © The Royal Society of Chemistry 2015 CrystEngComm Paper

Fig. 4 The crystal structures of (a) MT, (b) PA and (c) the MT–PA cocrystal.

Fig. 6 Temperature–composition phase diagram of the MT–PA Fig. 5 FTIR spectra of (a) MT, (b) the MT–PA cocrystal and (c) PA. system.

This journal is © The Royal Society of Chemistry 2015 CrystEngComm,2015,17,612–620 | 617 Paper CrystEngComm eutectics.35 The diagram shows a congruent melting temperature of 80.6 °C at a mole fraction of 0.5 which is the melting point of the MT–PA cocrystal and below those of MT (117.3 °C) and PA (107.1 °C). The two eutectics occurred at 71.1 °C and 78.9 °C, corresponding to 0.37 and 0.65 mole fractions of MT, respectively. The formation properties of a cocrystal AB are defined in reference to eqn (1):

A+B→ AB (1) where A, B, and AB are the crystals of component A (MT) and component B (PA) and the cocrystal of A and B (MT–PA), respectively. The formation enthalpy of the cocrystal can be determined by eqn (2),19,36

Δ Δ → − Δ → Hf = Hm(mix)(TS TL) Hm(cocrystal)(TS TL) (2)

– where TS is the temperature at which the MT PA cocrystal and the physical mixture of MT and PA are solid, TL is the temperature at which the cocrystal and the physical mixture Δ → both melt to the same liquid, and HmĲcocrystal) (TS TL) and ΔHmĲmix) (TS → TL) are the corresponding enthalpies of melting. Table 5 summarizes the key thermodynamic data. Fig. 7 Δ illustrates the determination of the formation enthalpies Hf of the MT–PA cocrystal. To obtain ΔHf, we integrated the heat-flow data for the MT–PA cocrystal and the corresponding physical mixture of the component crystals (Fig. 7) from a common liquid-state temperature (130 °C) down to a com- Fig. 7 (a) DSC melting endotherms of MT, PA, the MT–PA cocrystal and the 1 : 1 MT/PA physical mixture. (b) Relative enthalpies of MT, PA, mon solid-state temperature (25 °C). The enthalpy of the the MT–PA cocrystal and the 1 : 1 MT/PA physical mixture. cocrystal relative to the physical mixture is its formation −1 enthalpy. For this system, ΔHf is −1.19 J g at 50 °C (the reaction is exothermic), indicating that the cocrystal has a lower faster dissolution rate. The dissolution plot of the MT–PA enthalpy than the physical mixture of component crystals. cocrystal reached maximum solubility (Smax) within 20 min, This negative enthalpy change may also suggest stabilization which is followed by a decrease over time. This type of of the cocrystal compared to the components. “spring and parachute effect” has been exhibited by a lot of cocrystal systems recently.8,9 The peak MT concentration of the cocrystal is roughly double that of the API alone (Table 6). Dissolution studies The supersaturated solution is formed at the initial stage of The dissolution rate and apparent solubility of drugs are of dissolution and then preserved for several hours. Such a paramount importance in pharmaceutical development and behavior is especially favorable for pharmaceutical applica- quality control, and shorter dissolution times and higher tions. After 6 h, the concentration of the MT–PA cocrystal is apparent solubility may result in greater absorption. The close to that of pure MT, indicating that the cocrystal slowly powder dissolution profiles of MT, the MT–PA cocrystal and converts to MT accompanying the dissolution process, which the 1 : 1 MT/PA physical mixture in PBS (pH 6.8) are shown in was further confirmed by PXRD analysis of the remaining Fig. 8a in the form of solution concentrations of MT solids after the dissolution tests. − (mg mL 1) vs. time (min). Compared with intact MT and the Intrinsic dissolution rate (IDR) experiments were physical mixture, the cocrystal showed higher solubility and a performed in order to obtain quantitative information on the dissolution rates.37 The IDR results are graphed in Fig. 8b. Table 5 Key thermodynamic data for MT, PA, and the MT–PA cocrystal All of the samples demonstrated excellent linearity over the entire time span (60 min). The values obtained by least- a ° Δ −1 Δ b −1 Sample MP / C Hm/J g Hf /J g squares analysis of the linear regions of the dissolution pro- MT 107.1 106.0 0 files are provided in Table 6. From these results an intrinsic − − PA 117.3 172.5 0 dissolution rate of 0.075 mg cm 2 min 1 was obtained for the MT–PA 80.6 131.8 −1.19 cocrystal, about 30% higher than that obtained for pure MT a b Δ ° −2 −1 Melting point. Hf was evaluated at 50 C. (0.058 mg cm min ). In contrast, the physical mixture

Fig. 9 DVS isotherm plots of MT, the MT–PA cocrystal and the 1 : 1 MT/PA physical mixture at 25 °C.

increased solubility and dissolution rate that the cocrystal provides must be considered relative to this potential stability disadvantage. Dynamic vapor sorption is a well-established method for the measurements of water vapor sorption and desorption of drug solids. Consequently, a DVS study was performed to investigate potential physical stability concerns. The dynamic vapor sorption and desorption isotherms of MT, the MT–PA cocrystal and the 1 : 1 MT/PA physical mix-

Fig. 8 (a) Powder and (b) intrinsic dissolution profiles of MT, the ture are shown in Fig. 9. It was evident that at all RH levels, MT–PA cocrystal, and the 1 : 1 MT/PA physical mixture in PBS (pH 6.8) the MT–PA cocrystal sorbed more water than the original MT at 37 °C. but sorbed less water than the physical mixture. The cocrystal is considered nonhygroscopic as it sorbs less than 0.08% water even at high humidity (90% RH) through sorption and desorption cycles. Further, the sorption–desorption curves Table 6 Solubility and intrinsic dissolution rates of MT, the MT–PA are closed, suggesting that there was no solid state transfor- cocrystal, and the 1 : 1 MT/PA physical mixture at 37 °CinPBS(pH6.8) mation of the cocrystal under the experimental conditions. − − − Sample Solubilitya/mg mL 1 IDR/mg cm 2 min 1 Although the tendency of the MT–PA cocrystal to convert to MT when exposed to water is not a desirable property from a MT 0.92 0.058 MT–PA 1.93 0.075 processing point of view, the cocrystal is stable as an isolated MT–PA mixture 0.94 0.061 solid and its kinetic stability at 0–90% RH levels is sufficient to allow standard processing steps to be carried out. a Maximum solubility after 20 min. Conclusions shows a comparable intrinsic dissolution rate to that of MT. A 1 : 1 cocrystal of MT with PA was obtained using a melt This confirms the potential of this new cocrystal form to crystallization method by controlling the crystallization envisage new, more efficient formulations of MT. PXRD anal- between 50 and 70 °C. Both crystal structure analysis and ysis of the solids recovered after 60 min revealed that no FTIR spectroscopy revealed that the hydrogen bonding inter- phase transitions or crystallinity changes occurred during the actions occur between the amide group of MT and a carbox- intrinsic dissolution tests. ylic acid group of PA in the cocrystal, and the formed hydrogen bonds are weaker than those in the component phases. The binary phase diagram established by DSC confirmed DVS studies the ability of MT and PA to cocrystallize. The cocrystal melts Since the MT–PA cocrystal converted to MT after 6 h of expo- at T = 80.6 °C and the eutectic mixtures were identified at sure to liquid water during the powder dissolution experi- T = 71.1 °C, χMT = ~0.37, and T = 78.9 °C, χMT = ~0.65. The − ments, the disproportionation may also give rise to solid- formation enthalpy of the cocrystal was −1.19 J g 1 at 50 °C, indi- state physical stability concerns. Thus the advantage of cating an enthalpic stabilization effect on the cocrystallization.

The DVS measurements confirmed that the cocrystal is 15 N. Takata, K. Shiraki, R. Takano, Y. Hayashi and K. Terada, nonhygroscopic at 0–90% RH levels as an isolated crystalline Cryst. Growth Des., 2008, 8, 3032–3037. solid. For the cocrystal an approximate 2-fold increase in sol- 16 N. H. Chun, I. C. Wang, M. J. Lee, Y. T. Jung, S. Lee, ubility and a 30% increase in the intrinsic dissolution rate W. S. Kim and G. J. Choi, Eur. J. Pharm. Biopharm., 2013, 85, relative to MT indicate the potential of this new cocrystal 854–861. form to envisage new, more efficient formulations of MT. 17 I. C. Wang, M. J. Lee and S. J. Sim, Int. J. Pharm., 2013, 450, 311–322. Acknowledgements 18 N. Shan, F. Toda and W. Jones, Chem. Commun., 2002, 2372–2373. Financial support from the NSFC (grant no. 21101173, 19 S. Zhang, M. T. Harasimowicz, M. M. de Villiers and L. Yu, 91127002 and 21331007), NSF of Guangdong Province J. Am. Chem. Soc., 2013, 135, 18981–18989. (S2012030006240), and Guangzhou Pearl River New Star Fund 20 Y. Yan, J. M. Chen, N. Geng and T. B. Lu, Cryst. Growth Des., Science and Technology Planning Project (2013J2200054) is 2012, 12, 2226–2233. gratefully acknowledged. 21 D. J. Berry, C. C. Seaton, W. Clegg, R. S. Harrington, S. J. Coles, P. N. Horton, M. B. Hursthouse, R. Storey, References W. Jones, T. Friščić and N. Blagden, Cryst. Growth Des., 2008, 8, 1697–1712. 1 Ö. Almarsson and M. J. Zaworotko, Chem. Commun., 2004, 22 K. Fucke, S. Myz, T. Shakhtshneider, E. Boldyreva and 1889–1896. U. Griesser, New J. Chem., 2012, 36, 1969–1977. 2 P. Vishweshwar, J. A. McMahon, J. A. Bis and 23 D. X. Tan, L. C. Manchester, R. Hardeland, L. Silvia, M. J. Zaworotko, J. Pharm. Sci., 2006, 95, 499–516. C. M. Juan, M. S. Rosa and J. R. Russel, J. Pineal Res., 3 Guidance for Industry: Regulatory Classification of 2003, 34,75–78. Pharmaceutical Co-Crystals. U.S. Department of Health and 24 H. Peng, F. Bouak, O. Vartanian and B. Cheung, Int. J. Human Services Food and Drug Administration Center for Pharm., 2013, 458, 156–168. Drug Evaluation and Research. April 2013. Available from: 25 R. J. Reiter, Best Pract. Res., Clin. Endocrinol. Metab., 2003, http://www.fda.gov/Drugs/GuidanceComplianceRegulatory 17, 273–285. Information/Guidances/default.htm. 26 N. Buscemi, B. Vandermeer, R. Pandya, N. Hooton, 4 C. B. Aakeröy, S. Forbes and J. Desper, J. Am. Chem. Soc., L. Tjosvold, L. Hartling, G. Baker, S. Vohra and T. Klassen, 2009, 131, 17048–17049. Melatonin for treatment of sleep disorders. Evidence Report/ 5 A. V. Trask, W. Motherwell and W. Jones, Int. J. Pharm., Technology Assessment No. 108. Canada, 2004. 2006, 320, 114–123. 27 Circadin, melatonin. European Public Assessment Report. 6 S. Karki, T. Friščić, L. Fábián, P. R. Laity, G. M. Day and European Medicines Agency, 2007. W. Jones, Adv. Mater., 2009, 21, 3905–3909. 28 M. Vlachou, A. Eikosipentaki and V. Xenogiorgis, Curr. Drug 7 Y. Yan, J. M. Chen and T. B. Lu, CrystEngComm, 2013, 15, Delivery, 2006, 3, 255–265. 6457–6460. 29 B. J. Lee, H. G. Choi, C. K. Kim, K. A. Parrot, J. W. Ayres and 8 J. M. Chen, Z. Z. Wang, C. B. Wu, S. Li and T. B. Lu, R. L. Sack, Arch. Pharmacal Res., 1997, 20, 560–565. CrystEngComm, 2012, 14, 6221–6229. 30 B. J. Lee, K. A. Parrot, J. W. Ayres and R. L. Sack, Int. J. 9 S. L. Childs, L. J. Chyall, J. T. Dunlap, V. N. Smolenskaya, Pharm., 1995, 124, 119–127. B. C. Stahly and G. P. Stahly, J. Am. Chem. Soc., 2004, 126, 31 C. B. Aakeröy and D. J. Salmon, CrystEngComm, 2005, 7, 13335–13342. 439–448. 10 M. B. Hickey, M. L. Peterson, L. A. Scoppettuolo, 32 N. Qiao, M. Li, W. Schlindwein, N. Malek, A. Davies and S. L. Morrisette, A. Vetter, H. Guzmán, J. F. Remenar, G. Trappitt, Int. J. Pharm., 2011, 1–2,1–11. Z. Zhang, M. D. Tawa, S. Haley, M. J. Zaworotko and 33 CrysAlisPro Version 1.171.35.19, Agilent Technologies Inc., Ö. Almarsson, Eur. J. Pharm. Biopharm., 2007, 67, 112–119. Santa Clara, CA, USA, 2011. 11 D. P. McNamara, S. L. Childs, J. Giordano, A. Iarriccio, 34 G. M. Sheldrick, SHELXS 97, Program for Crystal Structure J. Cassidy, M. S. Shet, R. Mannion, E. O'Donnell and A. Park, Refinement, University of Göttingen, Göttingen, 1997. Pharm. Res., 2006, 23, 1888–1897. 35 H. Yamashita, Y. Hirakura, M. Yuda, T. Teramura and 12 A. V. Trask, Mol. Pharmaceutics, 2007, 4, 301–309. K. Terada, Pharm. Res., 2013, 30,70–80. 13 D. R. Weyna, T. Shattock, P. Vishweshwar and 36 S. Zhang, I. A. Guzei, M. M. de Villiers, L. Yu and M. J. Zaworotko, Cryst. Growth Des., 2009, 9, 1106–1123. J. F. Krzyzaniak, Cryst. Growth Des., 2012, 12, 4090–4097. 14 N. Rodriguez-Hornedo, S. J. Nehru and K. F. Seefeldt, Mol. 37 D. J. W. Grant and T. Higuchi, Solubility behavior of organic Pharmaceutics, 2006, 3, 362–367. compounds, Wiley, New York, 1990, ch. XI.