Introduction

Utilization of noncovalent interactions for the development and designed self-assembly of individual molecules is an emerging frontier in research, and the formation of cocrystals has been previously explored in these fields. Pharmaceutical cocrystals are homogeneous solid phases containing an active pharmaceutical ingredient (API) and one or more neutral molecular components called a cocrystal coformer (CCF) in a crystal lattice with defined stoichiometry (1,2). Their design relies on the predictability of supramolecular interactions between the complementary functionalities of cocrystal components and results in heteromeric interactions over their homomeric counterparts (3). These heteromeric interactions are usually formed by hydrogen bonds, aromatic π-stacking, and van der Waals forces. Cocrystallization is a particularly attractive design strategy for solid API phases because it could modify some of the physicochemical properties of a candidate drug, such as solubility, dissolution rate, melting point, stability, and bioavailability (4). To date, at least 90 APIs have been targeted for pharmaceutical cocrystallization due to their undesirable solubility, stability, and inherent hydrogen bonding capability (5).

Myricetin (MYR), 3,3’,4’,5’,5,7-hexahydroxyflavone (Fig. 1), is a flavonol that is present in many fruits, vegetables, and herbs, such as grapes, red wine, onions, berries, vine tea, and sargentodoxa cuneata (6–8). Recently, MYR was reported to exhibit potent iron chelating capacity, antioxidant and free-radical scavenging activities (9,10), as a result, it could be active against carcinogens in the body thus preventing mutations, and has potential for the treatment of diabetes, diarrhea, and thrombosis (11). However, the extremely low aqueous solubility (2 μg/mL) has limited its application in medicine. To improve an API’s solubility or dissolution rate, salt formation or cocrystal formation is a common alternative approach, especially for acidic compounds like MYR, which contains 6 phenolic hydroxyl groups. Comparatively, MYR is not suitable for salt formation because MYR is a rather weak acid with a pKa of 6.63 and an unstable component in a basic environment (12), an appropriate salt former (a strong base) would aggravate MYR’s instability. When salt formation is not optimal, cocrystallization is an alternative method for improving the physicochemical properties of an API because it features the formation of a complex between neutral molecules (13). Furthermore, MYR has many competitive hydrogen bonding sites, i.e., donors and acceptors (6 hydroxyls and 1 carbonyl), within its molecular framework, which provides the possibility for cocrystal formation of MYR with the proper CCFs. In this study, we opted for several pharmaceutically acceptable CCFs with relatively suitable structures to prepare pharmaceutical cocrystals of MYR to improve its poor solubility.

Fig. 1
figure 1

Chemical structures of MYR (a) and CCFs: CAF (b), NIC (c), INM (d), CYA (e).

Full size image

Previous approaches for screening and preparing cocrystals have advanced from being empirically based to a more efficient and rational basis. Currently, various preparation methods have been developed to obtain pharmaceutical cocrystals, and the solution and grinding methods are the most commonly used techniques (2). Generally, the types of CCFs and solvent have been considered during the process of cocrystal formation using these two methods; however, the stoichiometric ratio of the API and CCF, which is a critical factor, was usually fixed in many studies. For example, Dennis et al. (14) screened different CCFs and solvents to prepare the cocrystals of an API called compound 1 at two fixed stoichiometries (1:1 and 1:2) using high throughput screening (HTS) with a 96-well plate. Consequently, only two cocrystals with known stoichiometries were obtained: 1-benzoic acid cocrystal (1:1) and 1-maleic acid cocrystal (1:1). Although the majority of cocrystals have been reported to form at given ratios (1:1 and 1:2), other stoichiometries of individual components are also possible, such as acetazolamide-2,3-dihydroxybenzoic acid (3:1) (15), naproxen-nicotinamide (2:1) (16), cis,cis-1,3,5-cyclohexanetricarboxylic acid-urea (1:3) (17), and triphenylphosphine oxide-hydroquinone (2:3) (18). Accordingly, when fixing the stoichiometric ratio of an API and CCF during the screening and preparing process, other stoichiometric ratios are not tested. To obtain the cocrystals efficiently with all possible ratios, ratio selection should be carefully tested.

Ternary mixes, such as the API-CCF-Solvent system, can easily be presented using a ternary phase diagram (TPD) in which any point represents a fixed composition of the three substances. Such phase information can provide insight into the cocrystallization process, stability domains of various phases, and potential dissolution pathways (19). In addition, TPD could serve as a guide in selecting the proper solvent or synthesis method for cocrystal formation and provide the basis for developing large-scale preparative procedures using solution crystallization (14). However, early methods for drawing the TPD involved allowing an API (CCF) and cocrystal to equilibrate in a solvent with varying CCF (API) ratios (20). Thus, the cocrystals are obtained before constructing its TPD in this method. These processes are tedious, and the native role of TPD could not be embodied such as directly guiding cocrystal preparation and scale-up when cocrystal is not quite available. Therefore, another problem that emerged in the study of cocrystals is how to construct the TPD before the cocrystals are successfully prepared, simultaneously apply the TPD for preparing pure powder cocrystals.

In this paper, the cocrystallization of MYR with four CCFs, caffeine (CAF), nicotinamide (NIC), isonicotinamide (INM), and 4-cyanopyridine (CYA) (Fig. 1), were investigated. Briefly, a streamlined strategy based on the principles of reaction cocrystallization (21) is discussed with the aim of preparing cocrystals without knowledge of the stoichiometric ratio and by simultaneously constructing their TPDs. Furthermore, the strategy was verified. The detailed physicochemical characterization, including differential scanning calorimetry (DSC), Fourier transform-infrared spectroscopy (FT-IR), powder X-ray diffraction (PXRD), scanning electron microscopy (SEM), and nuclear magnetic resonance (NMR), are also presented and discussed with an emphasis on the future development potential of the four cocrystals. Lastly, the dissolution of MYR in the four cocrystals and pure MYR was evaluated.

Materials and Methods

Materials

The raw material (MYR) was purchased from Shanghai DND Pharm-Technology Co., Inc. (Shanghai, China). CAF, NIC, INM, and CYA were obtained from Sigma Aldrich Co. LLC (Shanghai, China). Methanol was purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). All chemicals were used as received without further purification.

Phase Solubility Diagram

The phase solubility of the MYR-CCF-solvent was determined by adding excess MYR to different concentrations of solutions of CCF while adding excess CCF to MYR. The resulting suspensions were allowed to reach equilibrium while stirring at 25°C. Aliquots of solutions were withdrawn after 12 h, centrifuged (13,000 rpm, 10 min), and the concentration of MYR and CCF in the supernatant were separately determined by HPLC (12).

Preparation of Cocrystals

In the pre-study, many candidate CCFs such as saccharin, caffeine, nicotinamide, isonicotinamide, 4-cyanopyridine, lysine, histidine, arginine, etc., were considered for MYR cocrystal preparation according to the cocrystal forming principle and cambridge structural database system. Eventually, only caffeine, nicotinamide, isonicotinamide, and 4-cyanopyridine could form pharmaceutical cocrystals with MYR. The detailed process for cocrystal preparation was as follows.

  • MYR-CAF. CAF (33.2 mg, 0.4% (w/w)) was dissolved in 10 mL of methanol (95.5% (w/w)). After, MYR (340.0 mg, 4.1% (w/w)) was added, and the mixture was stirred at 25°C for 12 h. The precipitate was obtained by suction filtration and dried in a fume hood for 24 h at room temperature. The other three cocrystals were prepared at the same temperature, feeding process, and post-processing except for the percentages of API, CCF, and solvent.

  • MYR-NIC. MYR (173.5 mg, 2.0% (w/w)), NIC (581.2 mg, 6.7% (w/w)), methanol (10 mL, 91.3% (w/w)).

  • MYR-INM. MYR (284.1 mg, 3.3% (w/w)), INM (404.6 mg, 4.7% (w/w)), methanol (10 mL, 92.0% (w/w)).

  • MYR-CYA. MYR (317.1 mg, 3.7% (w/w)), CYA (334.2 mg, 3.9% (w/w)), methanol (10 mL, 92.4% (w/w)).

Differential Scanning Calorimetry (DSC)

A differential scanning calorimeter (822e, DSC instruments, Mettler Toledo, Switzerland) was used to perform thermal analysis on the samples of MYR, CCFs, and MYR cocrystals. Powder samples of approximately 3.0 mg were placed in open aluminum crucibles and heated at a rate of 10°C/min up to 400°C.

Powder X-Ray Diffraction (PXRD)

The PXRD patterns of MYR, CCFs, and MYR cocrystals were obtained at ambient temperature using a Shimadzu XRD-6000X (Shimadzu, Japan). The samples were irradiated with Ni-filtered Cu-K (α) radiation at a voltage of 40.0 kV and a current of 40.0 mA. The scanning rate was 2°/min over a diffraction angle of 2θ ranging from 3° to 50°.

Fourier Transform-Infrared Spectroscopy (FT-IR)

FT-IR spectra were recorded on a Nicolet FT-IR-R330 spectrophotometer (ThermoFisher Scientific, Massachusetts, USA) over a range of 4,000 cm−1 to 450 cm−1 with a resolution of 1 cm−1 and 64 scans. MYR, CCFs, and MYR cocrystals were prepared in KBr disks with a hydrostatic press at a force of 5.2 T/cm2 for 3 min.

Scanning Electron Microscopy (SEM)

Morphological evaluation was performed using SEM (Philips XL-30, Eindhoven, Holland). A small piece of double-sided adhesive tape was fixed onto an aluminum stub, and the MYR and MYR cocrystal powders were sprinkled and dispersed on the stub surface. Prior to examination, the samples were sputter coated with gold-palladium under argon atmosphere to render them electrically conductive.

Nuclear Magnetic Resonance (NMR) Spectroscopy

1H-NMR spectra were acquired on a Bruker Avance III 600 MHz spectrometer equipped with a 5-mm PABBO probe (Bruker Corporation, Fällanden, Switzerland) at 25°C. Typically, approximately 10 mg of MYR, CCFs, and MYR cocrystals were dissolved in a glass vial in 0.5 mL of DMSO-d6. After, the solution was transferred into an NMR tube. All data were acquired and processed using Bruker Topspin 3.0 software.

Transition Concentration (Ctr) Measurement (Verification)

The transition concentration or invariant point was validated using a previously reported method (22). This was achieved by adding excess cocrystals to a saturated MYR solution containing excess solid MYR or a saturated CCF solution with excess solid CCF. After equilibration for 12 h, the supernatants were withdrawn and quantified by HPLC as described in the “Phase Solubility Diagram” section, and the obtained solid phases were characterized using DSC or PXRD. The verification experiments for the two Ctr values were performed in triplicate.

Powder Dissolution Experiments

For the dissolution studies of MYR and MYR cocrystals in powder, all samples were sieved through a 178-μm screen to produce powders with a similar range in particle sizes. The tests were performed on an RCZ-6C1 (Huanghai Medicine & Drug Testing Instruments, Shanghai, China) using the United States Pharmacopeia (USP) paddle apparatus II. The rotational speed of the paddle was set at 100 rpm and a temperature of the dissolution media was maintained at 25 ± 0.5°C. Accurately weighed powders corresponding to 50 mg of MYR were added to dissolution vessels containing 300 mL of medium. Three dissolution media were studied: 0.1 mol/L HCl (pH 1.2), 0.1 mol/L acetate buffer (pH 4.5), and 0.05 mol/L phosphate buffer (pH 6.8). Sampling was performed at 10, 20, 40, 60, 120, and 240 min, and the withdrawn slurry was filtered using 0.45-μm cellulose filters. A 20-μL sample was injected into an HPLC to quantify the concentration of MYR as mentioned in the “Phase Solubility Diagram” section. All tests were performed in triplicate.

Results and Discussion

Phase Solubility and Preparation of the Solid Phases

The phase solubility diagram (PSD) shows the concentrations of reactant in solution when different solid phases (API, cocrystal, and CCF) are at equilibrium, which was useful for determining the solution composition in specific phases and for studying solution complexation (23). Thus, we can ensure the formation of a new phase by observing the change in trends of the concentration curve in the PSD, which can also be used for directing the synthesis of cocrystals. The concentration of reactant in the PSD is the total concentration, which is originated from the unbound reactant and bound reactant, but it actually represents the saturation concentration at equilibrium. From this point, it is presumable that the cocrystal PSD could be constructed by adding an API and CCF to a proper solvent as long as the number of reaction ratios of the API and CCF was enough and that excess solid existed when the system was at equilibrium. Here, we used NIC as an example to explain our strategy for cocrystal preparation using the PSD. Considering the poor aqueous solubility of MYR and the preliminary experimental results (data not shown), methanol was selected as the solvent to prepare MYR cocrystals.

Figure 2 presents the PSD of the MYR-NIC-Methanol system, in which point a and b represent the solubility of MYR and NIC in neat methanol, respectively. From the obtained points in the PSD, the concentration of MYR decreased with the increase in NIC concentration when all phases reached equilibrium in methanol. The curve could be divided into three stages after all points were connected, in which the solid lines ac1 and c2b represent the solubility of MYR and NIC in different solutions, respectively. Furthermore, the trend of ac1 and c2b were relatively smoother compared to c1c2, which indicated that the solubility of MYR (ac1) and NIC (c2b) were less affected by the presence of the other component (NIC (ac1) or MYR (c2b)). However, as shown in line c1c2, the concentration of MYR (NIC) falls quickly when the concentration of NIC (MYR) is increased in the system, which indicated a new cocrystal might be formed, and c1c2 should represent its solubility (24). The intersected points c1 and c2 were called the transition concentrations (Ctr) or invariant points, and they represent the concentration of reactants in solution when the solid phase at equilibrium contains cocrystal and MYR (or NIC) (25). Therefore, the PSD refers to the solution composition of MYR and NIC at equilibrium with different solid phases.

Fig. 2
figure 2

PSD for the MYR-NIC-Methanol system at 25°C. The concentrations of reactant (MYR and NIC) in solution at equilibrium are shown in the PSD. “a” and “b” represent the solubility of MYR and NIC in methanol, respectively. “c1” and “c2” represent the transition concentration.

Full size image

However, the reactant concentration in solution is not enough to achieve cocrystal preparation. Unlike the PSD, the TPD shows the total composition of the system, including components in the solid and liquid phases (20,26). Thus, the phase solubility curve should be transformed into the TPD (Fig. 3). Furthermore, a straight line was drawn to connect these two points (c1, c2) and a new region appeared, which was labeled region 0. According to the principles of a TPD, if a MYR/NIC/Methanol mixture is generated with a starting composition in region 0, pure solid MYR-NIC cocrystal would be precipitated from the system, and the liquid composition can be varied on the curved full line c1c2. Obviously, region 0 is within the pure cocrystal forming zone, which cannot be obtained without knowledge of the cocrystal composition. However, region 0 could be easily found by drawing the PSD; therefore, it was useful to cocrystal preparation when the cocrystal composition is unknown (27–29).

Fig. 3
figure 3

The ternary phase diagram (TPD) was transformed from the PSD. The point in the TPD means the total amount of 100 g with MYR, CCF, and methanol. The crystallization of pure cocrystal forms in solution in region 0.

Full size image

We selected a point lying within region 0 of the TPD and created a mixture with the overall composition of 2.0% (w/w) MYR, 6.7% (w/w) NIC, and 91.3% (w/w) methanol to prepare MYR-NIC cocrystals as mentioned in the “Preparation of Cocrystals” section. Using the same method, the other three MYR cocrystals were also obtained.

Solid-State Properties

DSC

The DSC curves of MYR, CCFs, and cocrystals are shown in Fig. 4, from which it can be seen that pure MYR presented a single sharp endothermic melting peak at 356.06°C (Fig. 4a). CAF demonstrated a steep endothermic melting transition at 236.29°C (Fig. 4c), which is in agreement with the reported thermal behavior (30). Meanwhile, the DSC thermogram for MYR-CAF cocrystals showed a single endothermic transition, which was attributed to the melting transition (275.01°C) with a melting enthalpy of 142 J/g (Fig. 4b); the melting temperature was markedly different than the endothermic melting peak of MYR or CAF, suggesting the formation of a new phase. The MYR-NIC, MYR-INM, MYR-CYA cocrystals, and their CCFs resulted in similar phenomena with MYR-CAF cocrystals and CAF. They melted at 214.46°C, 257.53°C, and 191.51°C, respectively, with a melting enthalpy of 160 J/g, 59 J/g, and 232 J/g (Fig. 4d, f, and h). A single endothermic transition for the four cocrystals demonstrates that the new phases were stable below their melting points and indicates the absence of unbound or absorbed solvent and water (31).

Fig. 4
figure 4

DSC results of pure MYR (a), MYR-CAF cocrystal (b), pure CAF (c), MYR-NIC cocrystal (d), pure NIC (e), MYR-INM cocrystal (f), pure INM (g), MYR-CYA cocrystal (h), and pure CYA (i) with heating at 10°C/min.

Full size image

Furthermore, a direct correlation of chemical structure to the crystalline lattice energy and melting point is a challenging task because of several contributing factors, such as molecular arrangement in the lattice, noncovalent interactions, and conformational flexibility for a molecule (31), and this becomes more complex with cocrystals with multiple components. In the current study, an interesting pattern was found in that the melting points of all cocrystals were between that of MYR and the corresponding CCF, which is in agreement with the universal rule obtained from other cocrystals (32,33). It is clear that the melting point of the cocrystals was altered with the type of CCF. This information suggests that CCFs with higher melting points should be selected if MYR cocrystals with higher melting points are desired.

PXRD

The PXRD patterns of MYR, CCFs, and their cocrystals are shown in Fig. 5. The PXRD patterns of the four cocrystals clearly differed from those of the individual API and CCF, indicating the presence of a new crystalline phase. Specifically, several new reflections (marked with black triangles in Fig. 5) were observed; however, various characteristic reflections of MYR and the CCFs were absent. Taking the MYR-CAF cocrystals as an example, they exhibited some new characteristic reflections at 2θ 10.76°, 12.32°, 17.94°, 18.30°, and 21.94° (Fig. 5b), which allowed the MYR-CAF cocrystal to be distinguished from MYR and CAF. In addition, 2θ angles like 7.52°, 9.10°, 15.78°, 26.08°, 39.56°, 40.38°, 41.38°, and 46.62° of MYR (Fig. 5a) and 11.93°, 20.52°, 23.62°, 36.38°, and 44.26° of CAF (Fig. 5c) could not be observed in the powder pattern of the MYR-CAF cocrystals.

Fig. 5
figure 5

PXRD patterns of pure MYR (a), MYR-CAF cocrystal (b), pure CAF (c), MYR-NIC cocrystal (d), pure NIC (e), MYR-INM cocrystal (f), pure INM (g), MYR-CYA cocrystal (h), and pure CYA (i).

Full size image

FT-IR

IR spectroscopy is a reliable technique to characterize hydrogen bonding because a change in hydrogen bonds at the intermolecular level results in distinct vibrational frequency shifts.

As shown in Fig. 6a, MYR showed peaks at 3,415 cm−1 (corresponding to O-H stretching vibrations), 1,663 cm−1 and 1,618 cm−1 (-C = O stretching vibrations), 1,554 cm−1 (C = C stretching vibrations), 1,520 cm−1 (aromatic group), 1,326 cm−1 and 1,168 cm−1 (C–O–C vibrations), and 1,028 cm−1 (C–H bending vibrations). The four cocrystals (Fig. 6b, d, f, and h) showed distinct shifts in the hydroxyl and carbonyl region of MYR, which implied that the O-H group and C = O group in MYR were likely participating in the formation of hydrogen bonds in the cocrystallization process. For example, the O-H and C = O stretching frequency of the MYR-NIC cocrystal shifted from their original positions in MYR to 3,470 cm−1 and 1,661 cm−1, respectively.

Fig. 6
figure 6

FTIR results of pure MYR (a), MYR-CAF cocrystal (b), pure CAF (c), MYR-NIC cocrystal (d), pure NIC (e), MYR-INM cocrystal (f), pure INM (g), MYR-CYA cocrystal (h), and pure CYA (i).

Full size image

The four CCFs have the same chemical group, a N-containing heterocyclic ring, in which the N atom is a competitive acceptor (34). The C = N stretch shifts from 1,619, 1,624, 1,658, 1,593 cm−1 in NIC, INM, CAF, and CYA, respectively, to 1,606, 1,603, 1,646, 1,570 cm−1 in their corresponding cocrystals, which implies that the N-containing heterocyclic rings in the four CCFs participate in strong hydrogen bonds in the cocrystallization process. In addition, the C = O stretching vibrations shifted from 1,681, 1,664, and 1,700 cm−1 in NIC, INM, and CAF to larger wavenumbers (1,685, 1,686, and 1,705 cm−1) in their corresponding cocrystals, indicating that the carbonyl stretching vibrations of the amide groups was another characteristic for the formation of hydrogen bonds in MYR-NIC, MYR-INM, and MYR-CAF cocrystals. The phenomena also means that higher energies are required for stretching the C = O bonds when the CCFs are embedded in cocrystals (35). Differently, the IR vibrational modes corresponding to –CN in CYA were changed from 2,242 to 2,237 cm−1 in the MYR-CYA cocrystals, indicating the formation of a new supramolecular synthon –CN…H-O- in the cocrystals (36).

Thus, the variations obtained from comparing the spectra of the four MYR cocrystals with those of the starting components suggest that the cocrystals of MYR and the CCFs were formed with various hydrogen bonds.

SEM

The SEM results of MYR and its four cocrystals are presented in Fig. 7. As shown in Fig. 7a, the particle size of MYR is approximately 50 μm and its shape is always needle-like. However, the MYR-CAF, MYR-NIC, and MYR-INM cocrystals showed a smaller particle size ranging from 2 to 10 μm (Fig. 7b, c, and d). In addition, the MYR-CAF and MYR-NIC cocrystals exhibited a clear rod-like morphology, whereas the MYR-INM cocrystals exhibited block particles with a smooth surface. However, as shown in Fig. 7e, the MYR-CYA cocrystals crystallized in large block particles with many crystal features, such as rough surfaces and irregular edges. The diameter of the blocks ranged from 10 to 100 μm. The morphology of the particles revealed marked differences between MYR and the four synthesized cocrystals, which indicates that different crystal forms have been obtained.

Fig. 7
figure 7

SEM results of MYR and cocrystals. The differences between MYR and the four cocrystals in crystal quality (shape, surface, and size) are shown in pure MYR (a), MYR-CAF cocrystal (b), MYR-NIC cocrystal (c), MYR-INM cocrystal (d), and MYR-CYA cocrystal (e).

Full size image

NMR

Solution 1H-NMR was used to identify and confirm the ratio of MYR and the coformer in each cocrystal system. Three different stoichiometric cocrystals were formed, MYR-CAF (1:1), MYR-NIC (1:2), MYR-INM (1:2), and MYR-CYA (1:3) (Fig. 8), according to the ratio of the characteristic peak area of the API and CCF in the cocrystals (37).

Fig. 8
figure 8

Solution 1H-NMR patterns of samples to identify and confirm the ratio of MYR and CCFs in each cocrystal system: (a) pure MYR, (b) MYR-CAF cocrystal, (c) pure CAF, (d) MYR-NIC cocrystal, (e) pure NIC, (f) MYR-INM cocrystal, (g) pure INM, (h) MYR-CYA cocrystal, and (i) pure CYA.

Full size image

TPD and Verification

As previously stated, MYR and NIC crystallized with a 1:2 molar ratio in the MYR-NIC cocrystal. Converting this mole fraction to a mass fraction, a new point was obtained on the horizontal axis of the TPD (Fig. 9), which was labeled as point d, and this point is a constant value and does not change with solution composition (14). Point a is the solubility of MYR and point b is the solubility of NIC in neat methanol, and the values are 5.4% (w/w) and 20.2% (w/w), respectively. The c1d, c2d, c1MYR, and c2NIC points were connected, which meant the intact TPD of the MYR-NIC-Methanol system at 25°C was completed. In the TPD shown in Fig. 9, region 1 is the solution phase where the system is under-saturated with no solid at equilibrium. Regions 2 and 6 are the stable regions of pure solid MYR and NIC, respectively. The presence of NIC in methanol decreases the solubility of MYR to 5.0% (w/w) (ac1), which suggests a complex formation between the two components in dilute solutions (19). In contrast, the solubility of NIC always fluctuates around 20% (w/w) by the addition of MYR, which can be more clearly observed in the PSD (Fig. 2) and implies little change in the relative strengths of the intermolecular interactions. The reasonable explanation is that MYR is so weakly solvated by methanol (45 mg/mL) that the presence of a second solute with strong H-bonding ability would have a significant effect on its solubility. However, NIC is already solvated by methanol (201 mg/mL); therefore, its solubility is not greatly affected by the addition of MYR (38).

Fig. 9
figure 9

The ternary phase diagram of MYR-NIC-Methanol at 25°C in mass percent: (1) liquids, (2) MYR + liquid, (3) MYR + cocrystal + liquid, (4) cocrystal + liquid, (5) NIC + cocrystal + liquid, and (6) NIC + liquid. The black points were selected for TPD verification.

Full size image

In region 3, the solid phases of pure MYR and the 1:2 cocrystal are in equilibrium with a solution of invariant composition c1, whereas solid NIC and the cocrystal are at equilibrium with that of c2 in region 5. Pure 1:2 cocrystals in equilibrium with a solution of MYR and NIC is in region 4, from which we can clearly see the previous region 0 in Fig. 3 has been expanded to region 4; therefore, our strategy for preparing pure cocrystals is credible. However, these viewpoints are based on the precondition that our constructed TPD is exactly correct. Therefore, we have made an effort to prove the rationality of the TPD in the following works.

On one hand, the most direct way was performed by choosing the point in the drawn TPD randomly and then characterizing the solid phases after the reaction. Figure 9 shows the position of the selected points, and Figs. 10 and 11 separately give the DSC and PXRD analysis of the solid phases obtained from the different experimental conditions at equilibrium. Samples I, II, and III were prepared using ratios of API to CCF to solvent that correspond to a region of the TPD where at equilibrium both cocrystal and NIC would be expected to be present as solid phases. DSC analysis of solid phases I, II, and III (Fig. 10) demonstrated that these samples did indeed contain both cocrystal and NIC as melting events were observed at about 130°C and 215°C, respectively. However, the second peaks in solid phases I and II did not match the temperature of melting of cocrystal (214.46°C) in Fig. 4d and the similar situation happened in the curves of solid phases VII and VII, which should be the mixture of cocrystal and MYR. The probable explanation might be that the DSC process of a physical mixture like cocrystal with API or CCF was comparatively complicated so that some characteristic peaks would shift with the change of the ratios of cocrystal and CCF/MYR in the mixture or the cocrystal formation kinetics (39). Fortunately, the results of PXRD patterns of solid phases I, II, III, VII, and VII demonstrated that these samples did contain both cocrystal and NIC or MYR because both of their characteristic reflections (marked with red circles in Fig. 11) were observed. In addition, the DSC and PXRD patterns of the rest points (IV, V, VI, IX) perfectly matched with the predicted results from the TPD.

Fig. 10
figure 10

The DSC results of solid phases at equilibrium. The sample (I, II, III, IV, V, VI, VII, VII, IX) corresponds to the black points in the TPD, respectively.

Full size image
Fig. 11
figure 11

The PXRD results of solid phases at equilibrium and MYR-NIC cocrystal (a), pure NIC (b), pure MYR (c). The sample (I, II, III, IV, V, VI, VII, VII, IX) corresponds to the black points in the TPD, respectively.

Full size image

On the other hand, the Ctr was used for the verification because the two invariant points are critical indicators of cocrystal solubility and stability domains (40). First, the prediction of the phase transformations and determination of the thermodynamic stability of individual cocrystal systems are both closely related to the Ctr (22). Second, the Ctr can be used to determine cocrystal solubility, particularly for incongruently saturating cocrystals (metastable cocrystals) because the drug going into solution can be followed by crystallization in a supersaturable system (41–43). Lastly, knowing the critical points c1, c2, and d could accurately determine the main body of the critical region in the TPD (14), indicating that these three points are critical for the TPD construction. However, point d is fixed by the cocrystal composition. Thus, the other two points (c1 and c2) need to be validated using traditional methods. The results showed that two solid phases coexisted and the concentrations in solution (0.97 ± 0.08 mg/mL of MYR and 191.44 ± 0.44 mg/mL of NIC for c1, 35.64 ± 0.24 mg/mL of MYR and 41.08 ± 0.12 mg/mL of NIC for c2) were close to the predicted values (0.93 mg/mL of MYR and 191.32 mg/mL of NIC for c1, 36.18 mg/mL of MYR and 41.69 mg/mL of NIC for c2) at equilibrium, which confirmed that c1 and c2 were actually the Ctr points. These two experiments validated that our strategy for cocrystal formation was reasonable.

Here, we introduced a novel strategy for cocrystal formation. First, the PSD of MYR and CCF was constructed, from which the reactant concentrations in solution at equilibrium were obtained, and the change process of the phases was predicted. Second, by transforming the curve in the PSD to a TPD, a new region was obtained, which could guide us to prepare pure cocrystals. Third, after determining the composition of the prepared cocrystal by NMR, the TPD of the API-CCF-Solvent system was constructed.

The strategy has several advantages: (1) powder cocrystals could be prepared without knowledge of the stoichiometric ratio; therefore, this method could overcome the shortage of the fixed stoichiometric ratio in previous cocrystal research; (2) the TPD of the API-CCF-Solvent system was constructed by combining the constructed PSD and NMR data before the pure cocrystal was obtained, which was convenient and different from the traditional method; (3) cocrystal preparation and the TPD construction were simultaneously performed, indicating that this strategy was an efficient and economic measure for cocrystal research. However, similar to the other cocrystal research methods, it has its limitations. For example, the fact that the cocrystal of the drug and CCF could be definitively formed in the selected solvent was a precondition of this strategy. Furthermore, this strategy is not suitable for cocrystals with two or more stoichiometric ratios of API and CCF in the same solvent and temperature because there could be three or more invariant points under that condition, which would make it difficult to determine their accurate positions in the TPD. Therefore, to further optimize the strategy and simultaneously overcome these limitations as well as enlarge the method’s applicable scope is a critical issue and needs to be further investigated.

Powder Dissolution Studies

The powder dissolution profiles for the four cocrystals and pure MYR were obtained from the slurry experiments performed in three different pH buffers, which were used to evaluate the drug dissolution rate in different physiological environments (44). From the dissolution profiles (Fig. 12), the four cocrystals displayed an increased solubility and dissolution rate than MYR in each media; thus, the solubility of MYR in cocrystals was increased by approximately 3–80 fold and the time to reach the maximum value was decreased to only 10–20 min. In fact, the highest solubility of MYR exhibited by the MYR-CAF, MYR-NIC, MYR-INM, and MYR-CAY cocrystals in three media were enhanced from less than 1.5 μg/mL in the crude MYR to 11.9 μg/mL, 10.9 μg/mL, 9.8 μg/mL, and 7.1 μg/mL, respectively, which implies that cocrystallization is an effective approach for improving the solubility of MYR.

Fig. 12
figure 12

Powder dissolution profile of MYR and its four cocrystals in 0.1 M HCl aqueous solution (a), pH 4.5 acetate buffer (b), and pH 6.8 phosphate buffer (c).

Full size image

As shown in Fig. 12, the solubility of MYR in pH 4.5 acetate buffer is approximately 1.5 μg/mL, whereas it is nearly insoluble in 0.1 M HCl and pH 6.8 phosphate buffer, which suggests that the solubility of MYR is pH dependent in aqueous buffer. Similar to MYR, the four MYR cocrystals had increased solubility in pH 4.5 acetate buffer. All cocrystals exhibited higher solubility and reached maximum solubility more quickly than in the other two mediums. Here, the pH-dependent phenomenon of each MYR cocrystal might be dominated by the API (MYR), and the similar results were obtained in other cocrystals, such as carbamazepine-saccharin cocrystals (45), acyclovir-glutaric acid cocrystals (46), adefovir dipivoxil-saccharin cocrystals (44), and indomethacin-saccharin cocrystals (42).

However, in the same dissolution media, the solubility of a cocrystal might follow the CCF solubility rule (47); i.e., a high solubility CCF imparts a high solubility on the cocrystal. A good correlation between the solubility of the MYR cocrystals and their corresponding CCF in each media could be observed from Fig. 12 and SciFinder data. For example, the order of cocrystal solubility in pH 4.5 acetate buffer is: MYR-CAF > MYR-NIC > MYR-INM > MYR-CYA, which was consistent with the solubility order obtained from SciFinder, CAF (58 g/L) > NIC (53 g/L) > INM (45 g/L) > CYA (12 g/L). Thus, the CCF solubility rule was embodied in the four different MYR cocrystals. Other pharmaceutical cocrystals, such as pyrazinecarboxamide with various carboxylic acids and febuxostat with 5 pharmaceutically acceptable CCFs, followed the CCF solubility rule on their dissolution and solubility results (48,49).

Interestingly, several MYR cocrystals showed a slightly secondary peak concentration at approximately 120 min. The phenomenon might be explained by the dissociation of cocrystals into their components, which was attributed to weak intermolecular forces in the cocrystals (41). In general, during the dissolution process, the neat cocrystals transformed rapidly to the low-soluble drug (MYR), and MYR became supersaturated and immediately crystallized as the single-component MYR. Therefore, the high solubility of the four MYR cocrystals is not likely the maximum values, which were enhanced by the cocrystal technology. Supersaturation is a common phenomenon in many cocrystal dissolution experiments (40,50), which greatly confines the role of cocrystals in enhancing the solubility of insoluble drugs. Currently, the measures to solve the problem are mainly focused on the following two points: one is to calculate the solubility of cocrystals with the Ctr, which has been mentioned in the “TPD and Verification” section (41–43) and to choose the most suitable cocrystals from the results; the other method is to design a formulation of cocrystals with the aim of controlling the supersaturation levels (51). As far as the present study was concerned, whether the dissociation of cocrystals into the poorly soluble drug (MYR) and CCFs during dissolution could be avoided by the above mentioned approaches is unknown and worth future study.

In short, MYR cocrystals present a pH-dependent dissolution profile and exhibit a complete and rapid dissolution in all experimental aqueous media compared to pure MYR, which is probably because of crystal packing and the process of cocrystallization. From this research, it can be concluded that cocrystals provide a new approach for dissolution enhancement by forming intermolecular hydrogen bonding between MYR and CCF, which offers opportunities for MYR to be therapeutic agents.

Conclusion

Pharmaceutical cocrystals of MYR with CAF, NIC, INM, and CYA were successfully prepared and characterized by DSC, PXRD, IR, SEM, and NMR. A novel strategy for cocrystal preparation without knowledge of the stoichiometric ratio based on the TPD principle was developed. First, the PSD of MYR and CCF was constructed and was transformed into a TPD. Second, two invariant points were inferred and a part of the cocrystal forming region was obtained, which guided us in preparing pure cocrystal. Third, the TPD of the API-CCF-solvent system was completely constructed by determining the composition of the cocrystal by NMR. In addition, the dissolution results demonstrate that the solubility and dissolution rates of MYR were significantly improved by cocrystallization with the four CCFs, and all cocrystals exhibited pH-dependent solubility.