Solid-State Characterization and Relative Formation Enthalpies To

Apr 16, 2018 - Drug-Excipient Interactions: Effect on Molecular Mobility and Physical Stability of Ketoconazole–Organic Acid Coamorphous Systems. Mo...
0 downloads 4 Views 2MB Size
Download PDF
Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Solid-State Characterization and Relative Formation Enthalpies To Evaluate Stability of Cocrystals of an Antidiabetic Drug Naga Kiran Duggirala,†,∥ Heather L. Frericks Schmidt,‡ Zhaohui Lei,§ Michael J. Zaworotko,∥ Joseph F. Krzyzaniak,‡ and Kapildev K. Arora*,‡ †

Department of Pharmaceutics, University of Minnesota, Minneapolis, Minnesota 55455, United States Pfizer Worldwide Research and Development, Drug Product Design, Groton, Connecticut 06340, United States § Pfizer Worldwide Research and Development, Analytical Research and Development, Groton, Connecticut 06340, United States ∥ Department of Chemical Sciences and Bernal Institute, University of Limerick, Limerick V94 T9PX, Ireland ‡

S Supporting Information *

ABSTRACT: The current study integrates formation enthalpy and traditional slurry experiments to quickly assess the physical stability of cocrystal drug substance candidates for their potential to support drug development. Cocrystals of an antidiabetic drug (GKA) with nicotinamide (NMA), vanillic acid (VLA), and ethyl vanillin (EVL) were prepared and characterized by powder X-ray diffractometry (PXRD), spectroscopic, and thermal techniques. The formation enthalpies of the cocrystals, and their physical mixtures (GKA + coformer) were measured by the differential scanning calorimetry (DSC) method reported by Zhang et al. [Cryst. Growth Des. 2012, 12 (8), 4090−4097]. The experimentally measured differences in the relative formation enthalpies obtained by integrating the heat flow of each cocrystal against the respective physical mixture were correlated to the physical stability of the cocrystals in the solid state. The relative formation enthalpies of all of the cocrystals studied suggest that the cocrystals are not physically stable at room temperature versus their physical mixtures. To further address relative stability, the cocrystals were slurried in 30% v/v aqueous ethanol, and it was observed that all of the cocrystals revert to GKA within 48 h at room temperature. The slurry experiments are consistent with the relative instability of the cocrystals with respect to their physical mixtures suggested by the DSC results. KEYWORDS: pharmaceutical cocrystals, solid-state characterization, nuclear magnetic resonance, formation enthalpies, physical stability

1. INTRODUCTION

published guidance on cocrystals that suggest them to be explored as drug substances for new drug applications.17,18 Pharmaceutical cocrystals have been studied extensively in the past decade to customize the physicochemical properties of the API.4−16 Comprehensive efforts have been made to rationally understand the definition, design, diversity, development, and applications.4−16,19−28 However, to retain the advantages offered by cocrystals such as enhanced dissolution rate and solubility, the cocrystals should be physically stable during formulation, processing, and storage. In the drug development process, cocrystals would be mixed with a variety of excipients and exposed to different temperatures/humidity levels, and in some situations the stability of the cocrystal might be compromised; i.e., the cocrystal may undergo a phase change or dissociate to its individual components (API and coformers).29 Jones et al. have extensively studied the physical

A majority of medicines are formulated and administered as solid oral dosage forms such as tablets or capsules. Approximately 40% of marketed active pharmaceutical ingredients (APIs) and ∼70−90% of the drug candidates in research and development portfolios are classified as poorly water-soluble.1 Solubility enhancement of the API can be attained through salt formation, high energy metastable forms including amorphous solids, and enabling formulation such as complexation with cyclodextrins, utilizing lipidic systems, cosolvents, and micellar solutions.2 However, not all APIs are amenable to salt formation, and these high energy solid forms and formulations are associated with a risk of physical instability. For instance, amorphous APIs or metastable forms can exhibit an inherent propensity to crystallize to a lower energy form during manufacturing or storage.3 Cocrystallization of the API with acceptable coformers is becoming a more common approach to increase the solubility and stability of an API.4−16 Interestingly, the Food and Drug Administration (FDA) and European Medicine Agency (EMA) have recently © XXXX American Chemical Society

Received: Revised: Accepted: Published: A

January 18, 2018 April 12, 2018 April 16, 2018 April 16, 2018 DOI: 10.1021/acs.molpharmaceut.8b00061 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

reported by Zhang et al. The relative stability of the cocrystals was determined by solvent-mediated transformations. The pharmaceutical candidate (S)-6-{3-cyclopentyl-2-[4(trifluoromethyl)-1H-imidazol-1-yl]propanamido}nicotinic acid (referred as GKA throughout the paper) is a hepatoselective glucokinase activator proposed for the treatment of type 2 diabetes mellitus. Glucokinase catalyzes the phosphorylation of glucose to glucose-6-phosphate and functions as a key regulator of glucose homeostasis. The free form of GKA was isolated as an anhydrous crystal form and its relative stability assessed using computational methods described by Feeder et al.,38 which indicated that the molecular packing in the crystal structure was not optimal. Specifically, the strongest hydrogen bond donor and acceptor groups did not form hydrogen bonds, and a more thermodynamically stable crystal form was predicted (Supporting Information Figures S1 and S2; Table S1). An extensive polymorph screen was conducted to mitigate the risk of form conversion at a later stage of drug development. As a result, 12 new crystalline forms were identified including hydrates/solvates and other anhydrous forms of the free form (Table S2). These forms were found to be strongly solvent dependent, and it was therefore challenging to determine the thermodynamic relationships between the several crystalline forms. This complex polymorph landscape impeded the drug development process, and a salt screening was conducted to identify an alternate robust solid form. Most experiments resulted in noncrystalline films or oils, presumably due to rapid hydrolytic degradation of GKA under acidic and basic conditions. This was evident by the formation of the HCl salt of 6-amino-3-nicotinic acid while attempting salt formation with hydrochloric acid. Consequently, salt formation did not appear to be a viable approach to identifying a more robust form of the free form, and therefore cocrystallization with a suitable coformer was selected as a strategy for drug product development. Computational tools and Cambridge Structural Database (CSD) approaches39 were utilized to identify coformers with complementary functional groups for GKA, and five new cocrystals were isolated (Table S3). The cocrystals of GKA with nicotinamide (NMA), vanillic acid (VLA), and ethyl vanillin (EVL) were selected to measure the relative formation enthalpies based on their physical properties (melting point, crystallinity, and hygroscopicity). The structural formulas of the coformers used herein are shown in Chart 1.

stability of cocrystals in the presence of water vapor pressures and elevated temperatures.30 For example, the caffeine− theophylline cocrystal system dissociated to its individual components due to decrease in free energy at a higher temperature.30 In another study, the caffeine−adipic acid cocrystal system was found to partially dissociate after 10 days at 98% RH; in this case, the dissociation was induced by dissolution of adipic acid in water.31 Interestingly, Arhangelskis et al. showed dissociation for the pyrazine−phthalic acid cocrystal system at ambient conditions in less than 24 h, whereas other pyrazine cocrystals (with fumaric, succinic, and terephthalic acid) were indefinitely stable at ambient water vapor pressures but dissociated at higher relative humidity conditions.32 These reports suggest that dissociation of the inherently unstable molecular cocrystals is thermodynamically driven whereas stability of a solid form is kinetically controlled and dependent upon specific storage conditions. Therefore, the thermodynamic and kinetic stability of a cocrystal under conditions that they are likely to encounter in their usage is an aspect that needs to be thoroughly understood in order to develop cocrystals as active ingredients in a drug product. Recently two methods have been introduced to understand the thermodynamic stability (formation enthalpies) of cocrystals versus physical mixtures of their cocrystal components via (1) isothermal solution calorimetry and (2) differential scanning calorimetry. Oliveira et al. studied the enthalpy of formation using solution calorimetry for carbamazepine, cyheptamide, and 10,11-dihydrocarbamazepine cocrystals with saccharin as a coformer.33 The results suggested that these cocrystals were enthalpically favored, and the heats of formation were correlated to the hydrogen bonding changes of the molecular structures in crystal packing arrangements. Other examples of formation enthalpies measured using solution calorimetry include lamotrigine salts/cocrystals and felodipine cocrystals with 4,4-bipyridine.34,35 In the solution calorimetric method, the nonideal nature of solids (API or coformer) in solutions was taken into account by measuring the heat evolved from solution of one component (either API or coformer) in the presence of other and vice versa. Zhang et al. measured the relative formation enthalpies of cocrystals by DSC using nicotinamide−R-mandelic acid as model system.36 The method is based on the fact that the cocrystal and the physical mixture of cocrystal components produce the same liquid upon melting. In this case, the nonideal nature was corrected for by taking physical mixtures instead of individual components in the solid state. The authors demonstrated that the cocrystal has a lower formation enthalpy versus the physical mixtures. Further, Zhang et al. also used this method to calculate the formation enthalpies of nicotinamide−R-mandelic acid cocrystals with different stoichiometric ratios.37 Based on the available literature, the current research is focused on the fundamental question that if formation enthalpy of cocrystals can be determined through limited resources and material, can it be extrapolated to assess the stability of cocrystals of an API at an early stage of drug development process? This contribution addresses this question and is dedicated to two main objectives: (1) to determine the relative formation enthalpies of cocrystals with respect to the physical mixture of cocrystal components and (2) to evaluate the physical stability of cocrystals. The first objective was achieved by measuring the formation enthalpies of cocrystals with respect to the physical mixtures by using the DSC method

2. MATERIALS AND METHODS GKA was synthesized at Pfizer. The coformers (NMA, VLA, and EVL) were purchased from Sigma-Aldrich (purity 98%). All chemicals and solvents were used as received without further purification. The synthetic methods of preparation of Chart 1. Molecular Structures of (a) GKA, (b) Nicotinamide (NMA), (c) Vanillic Acid (VLA), and (d) Ethyl Vanillin (EVL)

B

DOI: 10.1021/acs.molpharmaceut.8b00061 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics cocrystals and the instrument methods and parameters are given in the Supporting Information.

3. RESULTS AND DISCUSSION 3.1. Screening and Design of Cocrystals. A search of the Cambridge Structural Database (CSD) was conducted to

Figure 3. Overlay of carbon ssNMR spectra of GKAEVL cocrystal, GKA + EVL physical mixture, EVL, and GKA (from top to bottom). Selected 1H T1 values of GKAEVL are shown.

Figure 1. Overlay of carbon ssNMR spectra of GKANMA cocrystal, GKA + NMA physical mixture, NMA, and GKA (from top to bottom). Selected 1H T1 values of GKANMA are shown.

Figure 4. Overlay of nitrogen ssNMR spectra. From top to bottom: coformer NMA, cocrystals GKANMA, GKAEVL, and GKAVLA, and GKA. The nitrogen numbering follows Chart 1.

Scheme 1. Schematic Representation of the Process To Integrate the Relative Formation Enthalpy between Cocrystal and Physically Mixed Cocrystal Componentsa Figure 2. Overlay of carbon ssNMR spectra of GKAVLA cocrystal, GKA + VLA physical mixture, VLA, and GKA (from top to bottom). Selected 1H T1 values of GKAVLA are shown.

identify the coformers with complementary functional groups that can serve as components for molecular recognition.39 Other computational tools such as COSMOquick and Mercury CFC 3.9, Molecular Complementarity Screening Module, were also utilized to select the appropriate coformers from a comprehensive coformer library for use in an extensive cocrystal screen.40,41 The COSMOquick tool is based on the miscibility of API and coformer in a supercooled liquid phase as measured by enthalpy of mixing. The complementarity module is conceptualized on comparison of molecular descriptors such as polarity, shape, and size between the API and coformer. The calculations and details can be found in Tables S4 and S5. Approximately 25 coformers showed high propensity to cocrystallize with GKA based on these three approaches (Table S3). Cocrystals of GKA with NMA (GKANMA), VLA

a

The process includes all endothermic and exothermic events.

(GKAVLA), and EVL (GKAEVL) were discovered during the screen and selected for further solid-state characterization and to measure the relative formation enthalpies. 3.2. Characterization of Cocrystals. The preliminary characterization of solid forms, GKANMA, GKAVLA, and GKAEVL, indicated formation of cocrystals with GKA to C

DOI: 10.1021/acs.molpharmaceut.8b00061 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

recorded for all signals. The resolved NMA peak at 149.5 ppm and GKA peak at 25.7 ppm had T1 values within this range (4.3 and 4.6 s, respectively, Figure 1). In the solid state, protons have very strong dipolar interactions, which result in the transfer of magnetization between nearby protons. This phenomenon, called spin diffusion, can occur at distances up to 50 nm in rigid systems.42 Components within the same crystalline lattice will undergo spin diffusion and have the same proton T1 value. This indicates that cocrystal formation had indeed occurred. The proton T1 experiment also provided evidence of cocrystal formation for GKAVLA and GKAEVL. For cocrystal GKAVLA, the overall T1 values were 2.5 ± 0.1 s with the VLA peak at 154.0 ppm having a T1 of 2.6 s and T1 of the GKA peak at 62.0 ppm being 2.6 s (Figure 2). For the GKAEVL cocrystal system an overall T1 of 4.3 ± 0.6 s was observed. The resolved EVL carbon at 64.6 ppm gave a proton T1 of 4.8 s. The resolved GKA peak at 61.5 ppm had a faster T1 value of 3.9 s, but this lies within the recorded error (Figure 3). In addition, all three cocrystals demonstrated changes in 1H T1 values from GKA, which had an average 1H T1 of 6.5 s (data not shown), further confirming cocrystallization of GKA with the coformers. Furthermore, as shown in Chart 1, the GKA molecule has two basic nitrogens: the imidazole (N2) with a predicted MoKa44,45 pKa of 3.04 and pyridine (N1) with a pKa of 2.91. In the GKAVLA complex the carboxylic acid of VLA coformer (predicted MoKa pKa of 4.32) could transfer its proton to these nitrogens. However, the pKa differences between GKA and VLA are negative, making this highly unlikely. GKA imidazole and VLA have a ΔpKa of −1.3, and GKA pyridine and VLA have a ΔpKa of −1.4. The NMA coformer has a weakly basic pyridine (N5) with a MoKa predicted pKa of 2.93 that can be ionized by the acid site of GKA (predicted pKa 3.68). The pKa differences between these sites are also negative (ΔpKa = −0.75). Furthermore, EVL has a predicted acidic pKa of 7.37 (−OH group), and the ΔpKa between GKA imidazole and EVL is −4.33 and GKA pyridine and EVL is −4.46. Therefore, GKAVLA, GKANMA, and GKAEVL are more likely to form a cocrystal.46 Moreover, the nitrogen ssNMR chemical shifts of the complexes were examined to verify they are indeed hydrogenbonded cocrystals and not salts (Figure 4). The nitrogen chemical shift differences between a complex and the free

Table 1. Average Formation Enthalpy of Cocrystals versus Physical Mixtures and Their Relative Formation Enthalpiesa solid component GKA + NMA physical mixture GKANMA cocrystal GKA + VLA physical mixture GKAVLA cocrystal GKA + EVL physical mixture GKAEVL cocrystal a

formation enthalpy (J/g) (ΔHmCCavg or ΔHmPMavg)

relative formation enthalpy (ΔHf) (J/g)

326.1 (5.9)

30.3 (5.9)

295.9 (0.7) 301.3 (2.5)

37.9 (2.8)

264.0 (1.0) 199.8 (7.3)

13.7 (7.8)

186.1 (2.0)

The formation enthalpy values are negative.

coformer stiochiometry of one to one (Figures S3−S5). To further confirm the cocrystal formation, solid-state nuclear magnetic resonance (ssNMR) spectroscopy was employed because it is sensitive to the crystal packing changes and other aspects of the local environment in a crystal lattice that would result from cocrystal formation.42,43 The analysis of data suggested noticeable changes in the carbon ssNMR spectra; for example, the aromatic carbons (105−160 ppm) of cocrystals exhibited large chemical shift differences in crystal lattice of cocrystals compared to the GKA and coformers. In contrast, the physical mixtures did not show any shifts in the individual components (GKA and coformers) (Figures 1, 2, and 3). In the cocrystal of GKAEVL, the change in the chemical shift of the EVL aldehyde from 190.5 ppm in the coformer to 194.9 ppm was observed when cocrystallized with GKA (Figure 3). EVL exhibits a strong intermolecular hydrogen bond between its aldehyde and phenol (PhOH···OC) groups. The hydrogen bond distance between the aldehyde and alcohol functional groups of EVL in cocrystal is either presumably longer or absent, causing the observed shielding decrease. Additionally, the EVL aldehyde moiety is expected to have a high propensity to interact with the carboxylic acid group (COOH···OC) of GKA. If this hydrogen bond is indeed present in GKAEVL, it would be expected to exhibit a longer hydrogen bond length as suggested by the ssNMR data. Additionally, 13C-detected proton T1 relaxation experiments were also performed on these cocrystals. For GKANMA cocrystal, an overall proton T1 value of 4.5 ± 0.5 s was

Figure 5. Relative formation enthalpies of cocrystals (green curve) versus physical mixtures (red curve) of GKA and coformers in 1:1 molar ratio as obtained by integrating the heat flow data from common solid state to the common liquid state: (a) from 20 to 214 °C for GKANMA; (b) from 20 to 210 °C for GKAVLA; (c) from 20 to 160 °C for GKAEVL. D

DOI: 10.1021/acs.molpharmaceut.8b00061 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 6. Powder X-ray diffraction patterns of cocrystals (a) GKANMA, (b) GKAVLA, and (c) GKAEVL, after slurrying all the three cocrystals in 30% v/v aqueous ethanol for 48 h. The characteristic GKA peak (4.9° 2θ) was observed for all three cocrystals as highlighted.

species can determine protonation versus hydrogen bonding.47 GKA imidazole (N2) and pyridine (N1) have nitrogen shifts at 252.6 ppm and at 260.7 and 262.7 ppm, respectively. (Two N1 peaks are a result of two GKA conformations being present in the crystal lattice.) The N1 and N2 sites of the GKAVLA cocrystal show minimal changes from free GKA. The shift differences are less than 1 ppm. If salt formation had occurred, changes of 50 ppm or higher would have been seen; therefore,

GKAVLA is proven to be a cocrystal. This approach is also validated in the GKAEVL cocrystal. The GKAEVL nitrogen shifts differ from GKA by less than 10 ppm. In addition, to confirm the N5 pyridine in NMA is not protonated in GKANMA, its nitrogen shifts were compared against free NMA. Free NMA has a single N5 peak at 304.4 ppm. In GKANMA two conformations are present with N5 shifts at 283.4 and 285.6 ppm. Differences between the free E

DOI: 10.1021/acs.molpharmaceut.8b00061 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics ΔHsolution = ΔHfusion + ΔHmix

coformer and the complex are less than 50 ppm, proving that GKANMA is a hydrogen-bonded cocrystal. A complete comparison of nitrogen shifts between GKA and cocrystals is provided in Table S6. 3.3. Relative Formation Enthalpies of Cocrystals. The main aim of this study was to understand the thermodynamic stability of cocrystals versus physically mixed GKA and coformers using the differential scanning calorimetry (DSC) method.36 The DSC curves of cocrystals, physical mixtures, and pure components are provided in Figure S6. In this study, when we refer to a physically unstable cocrystal, it suggests that it is dissociating to individual components as represented by eq 1. cocrystal(solid) → API(solid) + coformer(solid)

In solution calorimetry an ideal solution is achieved by measuring the formation enthalpy of one component in the presence of the other component to minimize solvent−solute interactions and can be represented by eq 6 ΔHsolution = ΔHfusion

(1)

(2)

where ΔH and ΔS are change in enthalpy and entropy, respectively. Thus, by measuring the change in Gibbs free energy, one can determine the relative stability of the cocrystal with respect to its physical mixtures. The entropy change (ΔSf) accompanied by a phase change of solid materials to liquid phase can be obtained from formation enthalpy (ΔHm) and melting point (Tm) of the solids using eq 3.48 Δsf = ΔHm/Tm

(3)

ΔHf = ΔHmCCavg − ΔHmPMavg

In this study, the formation enthalpy for each solid compound (cocrystal and physical mixtures of GKA and coformers) was measured and integrated from common liquid state temperature (5 °C above the melting temperatures) to common solid state temperature (room temperature). Hot-stage microscopy was conducted to confirm the liquid state of physical mixtures above the melting point of the cocrystals. Representative images for the GKA + VLA physical mixture are provided in Figure S7. Note that in the literature it was reported that occasionally the entropy could control the cocrystal formation;43 in this contribution the study was limited to measure the relative formation enthalpy based on the assumption that entropy is negligible compared to enthalpy in the solid state at room temperature. To obtain the formation enthalpies, the enthalpy required to convert solid components to its corresponding liquids for both cocrystals and physical mixtures was determined. The entire process may consist of several thermal events from initial solid components to final melt, for example, eutectic melting, dissolution of one component in second component (GKA in coformer, or vice versa). However, the resultant formation enthalpy obtained after the integration of heat flow from initial point (solid state temperature, TS) to final point (liquid state temperature, TL) is independent of the path as stated by Hess’s law and represented in Scheme 1.36 The formation enthalpies of a cocrystal and physical mixtures in the process can be represented by eq 4: ΔHm = ΔHexothermic events + ΔHendothermic events

(ΔHmix = 0)

(6)

In the DSC method we use herein, there is no solvent involved, and the formation enthalpy of each cocrystal and physical mixture can be measured by integration of heat flow of the cocrystals from common solid state to common liquid state temperatures provided the purity of API is same in cocrystal and physical mixtures (at least +5 °C above the melting temperature). The chemical purity data on cocrystals and physical mixtures after the DSC experiments are provided in Table S7. The formation enthalpies of the cocrystals were calculated by averaging ΔHmCC1, ΔHmCC2, and ΔHmCC3, i.e., ΔHmCCavg. The average formation enthalpy values of physical mixtures were calculated by averaging ΔHmPM1, ΔHmPM2, and ΔHmPM2 and are represented as ΔHmPMavg. ΔHmCCavg and ΔHmPMavg are the average formation enthalpies of each cocrystal and the physical mixtures of coformers at TS compared to the same liquid states at TL. Finally, the relative formation enthalpies of cocrystals, i.e., the enthalpy differences between cocrystals and physical mixtures, at TS, were calculated using eq 7.

The driving force for a phase change of a solid is Gibbs free energy (ΔG), represented in eq 2. ΔG = ΔH − T ΔS

(5)

(7)

The relative formation enthalpies for the three cocrystals were found to be 30.3 (5.9), 37.9 (2.8), and 13.7 (7.8) J/g for GKANMA, GKAVLA, and GKAEVL, respectively. Table 1 presents formation enthalpies (ΔHmCCavg or ΔHmPMavg) for three cocrystal systems. The relative formation enthalpies (ΔHf) as a function of temperature for GKANMA, GKAVLA, and GKAEVL are illustrated in Figures 5a, 5b, and 5c, respectively. The ΔHf is positive for each cocrystal, which suggests that the cocrystals are relatively unstable versus physical mixtures of the corresponding cocrystal components. 3.4. Stability Studies of Cocrystals. In early drug development, solution and suspension formulations are typically dosed in rodents, dogs, and nonhuman primates to achieve exposure during toxicology studies. To study the stability of cocrystals in aqueous solution, the cocrystal powder samples were slurried in water and the samples were subsequently analyzed by PXRD. The driving force for the dissociation or phase change of a cocrystal is the change in free energy (ΔG) at a particular temperature and pressure, but the dissociation process could be slow due to kinetic parameters or barriers. After 48 h, the PXRD data revealed no evidence of cocrystal dissociation in water. The stability of the cocrystal could be attributed to (1) negligible solubility of the cocrystal during aqueous slurry and/or (2) insignificant water sorption potential of cocrystals (