Predicting the Physical Stability of Amorphous Tenapanor

Jan 30, 2019 - The overall relaxation behavior which typically, is an indi-cator for physical stability, as measured by modulated temperature differen...
3 downloads 0 Views 3MB Size
Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX

pubs.acs.org/molecularpharmaceutics

Predicting the Physical Stability of Amorphous Tenapanor Hydrochloride Using Local Molecular Structure Analysis, Relaxation Time Constants, and Molecular Modeling Sanjeev Kothari and Radha R. Vippagunta* Pharmaceutical Chemistry and Formulations, Ardelyx, 34175 Ardenwood Blvd, Fremont, California 94555, United States

Mol. Pharmaceutics Downloaded from pubs.acs.org by UNIV OF GLASGOW on 02/15/19. For personal use only.

S Supporting Information *

ABSTRACT: The conformational flexibility of organic molecules introduces more structural options for crystallization to occur but has potential complications, such as, reduced crystallization tendency and conformational polymorphism. Although a variety of energetically similar conformers could be anticipated, it is extremely difficult to predict the crystal conformation for conformationally flexible molecules. The present study investigates differences in thermodynamic parameters for the free base, c-FB, and an amorphous dihydrochloride salt, a-Di-HCl, of a conformationally flexible drug substance, tenapanor (RDX5791). A variety of complementary techniques such as, thermal analysis, powder X-ray diffraction (PXRD), and molecular modeling were used to assess the thermodynamic properties and the propensity of crystallization for a-FB and a-Di-HCl, tenapanor. Molecular modeling and total scattering measurements suggested that the a-Di-HCl salt exists in an open elongated state with local 1D stacking, which extends only to the first nearest neighbor, while the a-FB shows local stacking extending to the third nearest neighbor. The overall relaxation behavior, which typically is an indicator for physical stability, as measured by modulated temperature differential scanning calorimetry and PXRD suggested a nontypical dual relaxation process for the dihydrochloride salt form. The first relaxation was fast and occurred on warming from the quench conditions without any thermal annealing, while the second relaxation step followed a more traditional glass relaxation model, exhibiting an infinite relaxation time. Similar analysis for the a-FB suggested a comparatively shorter relaxation time (about 19 days) that results in its rapid crystallization. This observation is further validated with the extensive amount of physical stability data collected for the a-Di-HCl salt form of tenapanor under accelerated and stress stability conditions, as well as long-term storage for more than 3 years that show no change in its amorphous state. KEYWORDS: conformational flexibility, tenapanor, amorphous, local molecular structuring, stability, relaxation time, density functional theory



INTRODUCTION

strategy is of great importance in an era of drug discovery where a large percentage of new molecules have solubility-limited dissolution rates.1−3 However, physicochemical processes leading to instability of amorphous drug products such as recrystallization and chemical degradation offset the potential benefit of greater aqueous solubility.4

Various solid forms of an active pharmaceutical ingredient (API) often display different mechanical, thermal, physical, and chemical properties that can remarkably influence its bioavailability, hygroscopicity, stability, and other performance characteristics. Hence, a thorough understanding of the relationship between the particular solid form of an API and its functional properties is important in selecting the most suitable form of the API for development into a drug product. Although a crystalline form of active compound is generally preferred, assessing the viability of an amorphous formulation © XXXX American Chemical Society

Received: Revised: Accepted: Published: A

August 9, 2018 January 25, 2019 January 30, 2019 January 30, 2019 DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 1. Chemical structure of tenapanor dihydrochloride.

combined with computational approaches can provide additional insight into the physical state of an amorphous solid. Materials that exhibit no local molecular order in the amorphous state have been observed to exhibit the minimum propensity for crystallization.24 In order for crystallization to occur, local molecular order must first develop. PXRD has been used on amorphous materials to probe the degree of local order and provide information on the propensity to crystallize.24 Especially, when crystal structure models are not available, individual molecular units can be derived using molecular structure minimization of a given molecule and can be tested against the measured total scattering signal to identify the most self-consistent molecular unit. Recent advances in solid-state modeling via density functional theory (DFT) have led to an increase in accuracy and efficiency in producing and predicting the most stable conformers.31 Once the most self-consistent molecular model is identified, lattice function can be derived. The lattice function defines the local, intermolecular packing in the noncrystalline state. The form of the lattice function is thus a sensitive probe of the inherent propensity for crystallization and is a precursor indicator of physical instability of the sample. The present study combines results from thermal analysis, PXRD, and molecular modeling along with predictions on conformational preferences to assess the propensity of the a-DiHCl to crystallize as compared to the a-FB.

Despite progress in recent years, the fundamental understanding and therefore the predictability of physical and chemical stability of amorphous phases is one of the main challenges in developing an amorphous drug product. Molecular mobility is generally thought to be a key factor governing the stability of amorphous phases and has been the subject of many studies.5−11 Although a detailed understanding of the relationship between mobility and stability of amorphous materials is still ongoing, especially below the glass transition temperature, Tg, there is enough evidence to support a strong connection between the molecular mobility and the stability of amorphous materials.12−18 However, structural complexity and the flexibility of the molecule, which manifests itself as an entropic barrier to crystallization,19−26 have rarely been discussed. We describe a study wherein a simple hydrochloride salt of a small molecule, tenapanor (RDX5791, a-Di-HCl) with a relatively large molecular weight (1197 Da), resists crystallization, although its free base form (c-FB) exists in a crystalline state. RDX5791 (tenapanor) [CAS name: (12,15-dioxa-2,7,9triazaheptadecanamide,17-[[[3-[(4S)-6,8-dichloro-1,2,3,4-tetrahydro-2-methyl-4-isoquinolinyl]phenyl]sulphonyl]amino]N-[2-[2-[2-[[[3-[(4S)-6,8-dichloro-1,2,3,4-tetrahydro-2-methyl-4-isoquinolinyl]phenyl]sulphonyl]amino]ethoxy]ethoxy]ethyl]-8-oxo-hydrochloride (1:2)] as shown in Figure 1, is currently in late-stage clinical trials for multiple indications. As the amorphous dihydrochloride salt was chosen for development due to better in vivo performance linked to better solubility and dissolution properties compared to the free base form, understanding its physical stability was of utmost importance. The current study attempts to assess the propensity of a-Di-HCl to crystallize. Many analytical techniques have been developed to monitor and investigate the recrystallization of amorphous drugs. These techniques include thermal analysis such as differential scanning calorimetry (DSC), solution calorimetry, powder X-ray diffraction (PXRD), and spectroscopic techniques, such as NMR (nuclear magnetic resonance), FT-IR (Fourier-transform infrared spectroscopy), NIR (near-infrared), and Raman spectroscopy.27−33 In general, thermal analysis techniques have proven to be very useful in assessing the bulk thermal properties of the solid sample associated with recrystallization processes. The average rate of molecular motions at any given temperature is probably the most important parameter to know for amorphous pharmaceutical materials, and it can be used to explain and even predict the stability of amorphous systems.25,27 It can be used to monitor the extent of relaxation at temperatures below glass transition (Tg), and heat capacity changes as a function of temperature.12−18 However, these thermal techniques are not able to give specific information related to molecular and crystal structures. PXRD measurements



EXPERIMENTAL SECTION Materials. Crystalline tenapanor free base (c-FB) and amorphous tenapanor dihydrochloride (a-Di-HCl) samples were used in this study. c-FB was converted into noncrystalline tenapanor free base (a-FB) by melt-quenching. Differential Scanning Calorimetry (DSC). DSC analyses were carried out for both, crystalline tenapanor free base (c-FB) and amorphous tenapanor dihydrochloride (a-Di-HCl) using a TA Instruments Q2000. The instrument temperature calibration was performed using indium. The sample was heated from ambient temperature to 350 °C at a rate of 10 °C per minute. The DSC cell was kept under nitrogen purge of about 50 mL per minute during each analysis. Modulated Differential Scanning Calorimetry (MDSC). The presence of significant volatile content contribution in the DSC heat flow data between 40 and 110 °C resulted in effectively hiding the glass transition event for a-Di-HCl and aFB obtained after a heat−cool−heat run of c-FB. To enhance the visibility of the glass transition, a-FB and a-Di-HCl were equilibrated at 170 °C, which is above the observed melting temperature of the crystalline free base form (150 °C). After equilibration within the DSC unit, the samples were rapidly cooled to −40 °C (to quench the noncrystalline state) and held isothermally for 15 min. The modulated DSC scans ran from B

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics −40 to 170 °C at a rate of 4 °C/min with a modulated amplitude of 0.8 °C and modulation period of 40 s. Differential Scanning Calorimetry (DSC)−Absolute Heat Capacity. To better determine the change in heat capacity between the noncrystalline forms a-FB and a-Di-HCl, a series of absolute heat capacity measurements were performed on TA Instruments Q2000. The absolute heat capacity determination follows the ASTM E1269−11 procedure and makes use of sapphire and empty pan heat flow measurements to correct the baseline and scale individual heat flow results for each sample. Due to the dominant volatiles contribution close to the expected glass transition event, absolute CP measurement for a-Di-HCl was performed by first equilibrating the a-Di-HCl at 100 °C for 15 min before quenching to −90 °C. The 100 °C equilibration step was designed to remove the volatiles without thermally stressing the salt. The c-FB was first equilibrated at 170 °C for 15 min before performing the absolute CP measurements. The equilibration temperature is above the free base melt (150 °C) and lower than the thermal degradation events. After equilibration, the liquid free base was quenched to −90 °C and held for 15 min before heating to the starting temperature of 5 °C. The quenched free base was held at 5 °C for 15 min before beginning the DSC scan from 5 to 200 °C at 10 °C per minute. Powder X-ray Diffraction. The Rigaku Smart-Lab X-ray diffraction system used for this study was configured as a Bragg− Brentano system using a line source X-ray beam. The X-ray source is a Cu Long Fine Focus tube that was operated at 40 kV and 44 mA. The data collection range was extended to 80° 2θ in order to support the normalization procedures used during molecular modeling. A reflection X-ray geometry with low background silicon sample holders was used to better control the variable background observed. The Bragg−Brentano configuration utilized is a clean optical system with the only active optical component being a beta filter. The D’teX detector used in this configuration was a position sensitive detector (PSD) that captures diffraction events over a relatively large diffraction range. Data Analysis: Methodology. The measured PXRD data files collected on the noncrystalline forms represent the total scattering signal from the sample. This data forms the basis of the total scattering modeling to characterize the local molecular structure within the noncrystalline state. Total scattering data analysis is an iterative procedure, wherein data preprocessing, and Debye calculations are performed together within each iterative step in the following sequence25 along with molecular modeling: (a) Data correction/preprocessing. (b) In parallel, DFT/molecular modeling is carried out by: (i) Generation of molecular models by DFT calculations. (ii) Use of the Debye−Menke total scattering approach to simulate the separate Debye and Menke contributions for selected molecular models. (iii) Combination of both single curves to simulate the Debye−Menke total scattering curve. (c) By combining the experimentally corrected XRPD pattern recovered at point “a” with the Debye−Menke contribution generated at point “b”, and by using of the formula given in next sections, the lattice function can be recovered. For this, an additional normalization step is necessary.

(d) Extraction of information about local order from the lattice function. (e) When a noncrystalline system becomes physically unstable and begins to undergo crystallization, the lattice function describing the arrangement of molecular units must begin to develop local intermolecular order. Total Scattering Modeling. Total scattering measurements are utilized in order to probe the local structure of noncrystalline systems. To extract the local structure from the observed data, computational modeling of the observed total scattering signal is required. Total scattering is usually modeled utilizing the Debye diffraction eq (eq 1)37 either in the form of a pair-wise distribution function (PDF) or through direct molecular modeling. Debye Equation n

IDebye =

∑ fi 2

n

+2

i=1



fi f j sin(Qdij)/(Qdij)

i , ji ≠ j1

(1)

This assumes an atomic level molecular model for the complete sample of interest, where n is the number of atoms in the model (with indices i and j), f i is the individual atomic form factor for atom i, and dij is the distance between atoms i and j. A more tractable approach to the calculation of the total scattering signal for a noncrystalline system is to use the lattice function formalism, which can be used with a single representative molecular unit, shown in eq 2. The lattice function approach assumes that the noncrystalline state can be described by a single average molecular unit that is arranged throughout the solid sample according to a random lattice function. Lattice Function Formalism Itotal diffraction = (IDebye − IMenke) × lattice function

(2)

where IMenke is the small angle scattering correction for the chosen molecular unit. Both the Debye equation and Menke correction can be calculated from a given molecular model.38 Menke correction is used to remove coherent SAXS response due to shape and mean number density. n

Menke correction =

∑ ∑ fi f j i=1 j=1

sin(Qlci) sin(Qlcj) Qlci Qlcj

Divisions of the measured data by the Debye models leave the lattice functions, which are parts of the diffraction responses resulting from molecular packing as per eq 3. i 1 y i 4π y i 2θ y Q jjjin zzz = jjj zzzsinjjj zzz k Å{ k λ { k 2 {

(3)

The data preprocessing is performed iteratively with the total scattering calculations to determine the most self-consistent molecular model. Experimental Data Preprocessing. In order to compare experimentally determined PXRD data for noncrystalline materials with computationally calculated data from the total scattering method, the measured data has to be rendered into an C

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics instrumental free form called a reduced structure factor. The preprocessing steps applied are (1) Instrumental background removal and digital filtering. (2) Correction for instrument intensity function (Lorentzpolarization + optics). (3) The normalization step is carried out to transform the Yaxis from measured intensity to electron units. (4) Removal of Compton scattering. (5) Debye normalization. To ensure self-consistency between the derived total scattering responses for these data files, an additional high angle normalization was performed. The normalization step transforms the Y-axis from measured intensity to electron units. With the data transformed into electron units, the Compton scattering correction can be performed. As such, this step of the data preprocessing is performed iteratively with the molecular modeling and total scattering calculations. With the removal of Compton scattering, the preprocessed data can be directly compared to the total scattering Debye−Menke response for the selected molecular model. This allows a final normalization step where the preprocessed data is scaled to give a best match to the Debye−Menke total scattering response for the selected molecular unit. This preprocessing step is again an iterative step performed in combination with molecular modeling and total scattering calculations. After Debye normalization, the lattice function can be derived based upon the total scattering Debye−Menke curve for the selected molecular unit model. This methodology was employed for all PXRD data files presented in this study. Molecular Modeling. To enable interpretation of the total scattering data for a noncrystalline system, development of molecular models that best describe the essential molecular unit underlying the noncrystalline state is required. Individual molecular units were derived using molecular structure minimization of a given molecule. Given the molecular structure of tenapanor dihydrochloride as depicted in Figure 1, a molecular model was constructed for both, the free base and the dihydrochloride salt using the Spartan molecular modeling package.

Figure 2. Modulated DSC analysis of the noncrystalline dihydrochloride (a-Di HCl) after equilibration at 100 °C.

Figure 3. Modulated DSC analysis of the noncrystalline free base (aFB) after equilibration at 170 °C.

Table 1. MDSC Results at Glass Transition Event



RESULTS AND DISCUSSION In order to broadly define a range of conditions under which the noncrystalline forms of a-Di HCl and a-FB were well behaved, the initial study was carried out at nonambient conditions. For a-Di HCl, the loss of volatiles under 150 °C interfered with observation of a glass transition (Tg) temperature in that region. Therefore, before analyses by MDSC, samples of a-Di HCl were equilibrated at 170 °C, as shown in Figure 2. a-FB samples analyzed by MDSC were prepared by equilibrating cFB at 170 °C, which is above the melting point of free base (150 °C), and quenching at −40 °C, as shown in Figure 3. The noncrystalline dihydrochloride (a-Di-HCl) has glass transition around 115 °C (Figure 2), which is significantly higher than that for the amorphous free base (a-FB), which is 60 °C (Figure 3). The absolute change in heat capacity (δCp) during each Tg event was obtained from the reversing heat flow curves, as shown in Table 1. These changes provide a relative measure of the molecular mobility in the glassy phase close to ambient conditions. a-FB exhibits a substantially larger δCp than does a-Di HCl, suggesting that a-FB is a more fragile glass that would be expected to rapidly develop local order upon cooling.

sample

Tg (°C) at center

δCp (J/g·°C)

H (J/g)

free base (a-FB) dihydrochloride salt (a-Di HCl)

60.03 115.48

0.3640 0.2835

2.174 2.453

Relaxation Time Constants. Relaxation time constants below Tg were determined using the method of Hancock,14 which utilizes aging experiments. In MDSC, the noncrystalline forms (a-FB and a-DiHCl) were freshly prepared in situ by heating each sample to 170 °C and holding them isothermally for 15 min. The samples were then rapidly cooled at −20 °C/ min to −90 °C where they were again held isothermally for 15 min. After cooling, each sample was slowly warmed at 4 °C/min to the target annealing temperature. The target annealing temperatures were selected to be 12, 24, and 48 °C below the glass transition event for each noncrystalline sample. A series of different annealing times were selected for each target temperature. For the initial study, all samples were annealed for 0, 2, 4, and 8 h. However, for the dihydrochloride salt, no relaxation was observed over these annealing times for the Tg-48 °C target temperature. For this specific Tg-48 °C temperature point, the samples were annealed for 0, 8, 16, and 32 h. The MDSC data obtained from the above samples were utilized to determine two experimental parameters: (1) absolute change in heat capacity (δCp) on passing the glass transition event and (2) relaxation enthalpy (δH) released. The individual δCp values are averaged to give a single number for the D

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

The extent of relaxation at each annealing point (ΦT) is determined using eq 5.

dihydrochloride salt a-Di HCl and the free base a-FB. The calculated mean δCp values are presented in Table 2.

ΦT = 1 − (δH /δH∞)

Table 2. Calculated Mean δCp Values for Free Base and Dihydrochloride Salt39 sample

δCp (J/g·°C)

free base (a-FB) dihydrochloride salt (a-Di HCl)

0.334 0.246

The extent of relaxation numbers presented in Table 6 gave numbers less than zero for the Tg-48 °C dihydrochloride annealing experiment. Table 6. Extent of Relaxation (ΦT) Determined for Each Annealing Experiment

Assuming ideal linear relaxation behavior, the δCp values (from Table 2) can be used to determine the maximum amount of relaxation enthalpy generated for infinite annealing times (δH∞) depending on the temperature below Tg at which the annealing was performed (Tanneal), using eq 4.

sample

sample

temperature

δH∞ (J/g)

dihydrochloride salt (a-Di HCl)

Tg-48 Tg-24 Tg-12 Tg-48 Tg-24 Tg-12

12.24 6.12 3.06 16.03 8.02 4.01

Tg-48 Tg-24 Tg-12 Tg-48 Tg-24 Tg-12

0.00 0.00 0.00 0.00 0.00 0.00

−0.12 1.38 2.11 0.13 1.16 2.57

−0.13 2.06 2.31 0.18 1.51 3.10

−0.28 2.62 2.68 0.04 2.20 3.78

sample dihydrochloride salt (a-Di HCl)

temperature

0h

2h

4h

8h

Tg-48 Tg-24 Tg-12 Tg-48 Tg-24 Tg-12

2.65 2.49 1.56 2.04 2.07 2.14

2.53 3.87 3.67 2.17 3.23 4.71

2.52 4.55 3.87 2.22 3.59 5.25

2.37 5.11 4.23 2.07 4.27 5.92

temperature

0h

8h

16 h

32 h

Tg-48

3.05

3.07

3.79

3.80

temperature

0h

8h

16 h

32 h

Tg-48

0.00

0.02

0.74

0.76

The extent of relaxation as a function of annealing time (t) was modeled using eq 6 to determine optimum values of the mean relaxation time constant (η) and relaxation time distribution parameter (β). ΦT = exp(− (t/η)β)

(6)

For small molecule systems in general, a typical value of the distribution parameter is around 0.5, with a practical range between 0.3 and 0.5 near Tg, and approaches 1.0 at high temperatures.14 The values for the distribution parameter and mean relaxation time constants for each annealing run are presented in Table 8. Table 8. Distribution Parameters and Time Constants Derived from Annealing Studies

Table 5. Extended Hold Time δH Values for Dihydrochloride Salt dihydrochloride salt (a-Di HCl)

8h

Table 7. Extent of Relaxation (ΦT) Determined for Longer Hold Time Experiments

Table 4. δH Values As Determined from Nonreversing Heat Flow Curves

sample

4h

This indicates no relaxation was occurring within the experiment time scale beyond the initially observed relaxation. For this reason, the more extended hold times were employed in a follow up measurement; as shown in Table 7.

The individual δH values as determined from the nonreversing heat flow curves are presented in Tables 4 and 5.

free base (a-FB)

2h

(4)

Table 3. Calculated δH∞ Values for Dihydrochloride and Free Base

dihydrochloride salt (a-Di HCl)

0h

dihydrochloride salt (a-Di HCl)

Calculated values for δH∞ as per eq 4 are shown in Table 3 for the dihydrochloride salt a-Di HCl and the free base a-FB for all annealing temperatures used.

sample

temperature

free base (a-FB)

δH∞ = (Tg − Tanneal)δCp

free base (a-FB)

(5)

distribution β

temperature

time constant (h)

dihydrochloride salt (a-Di HCl)

0.47

free base (a-FB)

0.65

Tg-48 Tg-24 Tg-12 Tg-48 Tg-24 Tg-12

11718 26.4 1.37 13462 44.3 1.96

sample

It was noted that even for the 0-h hold time experiments, a significant relaxation enthalpy was determined for both, a-Di HCl and a-FB. For the free base, this starting enthalpy was essentially constant for all runs (∼2.1 J/g). The starting enthalpy for the dihydrochloride salt varied for each of the annealing cycles. This type of behavior is not representative of “ideal” relaxation phenomena. The enthalpy values for the extended hold time experiments on the a-Di-HCl are shown in Table 5 and continued to show variability.

The relaxation time constants as a function of the annealing temperature can be described by the empirical Vogel−Fulcher− Tammann (VFT) stretched exponential eq (eq 7), where the pre-exponential factor η0 has been related to the single molecule relaxation times. In this equation, D* is the kinetic fragility parameter and T0 is a fictive temperature where the energetic barriers to relaxation effectively become infinite. The kinetic E

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics fragility parameter D* will typically take values between 2.0 and 100.0. For organic molecular glasses in general, D* is about 10.0.34 The fictive temperature T0 is generally approximated as Tg-50 K.34 η = η0exp(D*T0/(T − T0))

second relaxation step followed a more traditional glass relaxation model, exhibiting the long relaxation times, as predicted. Traditionally, relaxation times are critical indicators for physical stability. This, however, is based upon the existence of a crystalline form that exists with lower free energy than the relaxed glass. In several crystallization attempts, a stable dihydrochloride salt could not be successfully isolated as a crystalline solid.33 This is consistent with the long relaxation time for a-Di-HCl calculated from the VFT model. X-ray Powder Diffraction. The measured PXRD for the noncrystalline dihydrochloride (a-Di HCl), crystalline free base (c-FB), and amorphous free base (a-FB) are shown in Figure 6.

(7)

Excel solver was used to model the observed relaxation times with respect to eq 7 giving the output curve fits presented in Figures 4 and 5 for the a-Di-HCl and a-FB, respectively.

Figure 4. VFT modeling of relaxation time for a-Di-HCl.

Figure 6. Measured PXRD data for noncrystalline dihydrochloride (aDi HCl), crystalline free base (c-FB), and amorphous free base (a-FB).

The PXRD data obtained from a-FB and a-Di HCl were processed to provide lattice functions, which were used to assess the extent of local order in these materials. In order to compare the experimentally determined, PXRD data for noncrystalline materials with data calculated by the total scattering method, measured data were processed to remove instrumental contributions and normalized according to the procedure outlined in the experimental section. This data forms the basis for what is called total scattering modeling, which is used to characterize the local molecular structure within each noncrystalline material. Total scattering modeling is an iterative procedure with data preprocessing, molecular modeling, and Debye calculations being performed together within each iterative step. Normalization depended on the molecular model used. The first step was to generate a Debye model, which is simply the PXRD pattern expected from a single molecule in a single conformation (molecular unit), without consideration of packing. In order to describe the molecular unit underlying the noncrystalline state, various molecular units of tenapanor were derived using molecular structure minimization. In the absence of known crystal structures, energy-minimized, singlemolecule conformers represent the best choice for a molecular unit. Molecular models were defined for both “open” and “closed” conformers of tenapanor free base and tenapanor dihydrochloride, and the calculated total scattering curves were compared with the experimentally observed data. Different potential conformers were first isolated using a simple molecular dynamics model with the phenomenological MMFF (Merck Molecular Force Field). In an inert local environment (neutral like nitrogen atmosphere or vacuum), both tenapanor and tenapanor dihydrochloride were observed to fold up into a

Figure 5. VFT modeling of relaxation time for a-FB.

The derived VFT parameters for each material are shown in Table 9. Table 9. VFT Fitting Parameters sample

η0 (× 10−9 h)

T0 (K)

D*

dihydrochloride salt (a-Di HCl) free base (a-FB)

0.86 0.83

251.0 192.0

10.1 13.9

The VFT curve fitting resulted in fictive temperatures (T0) for both the free base and dihydrochloride at about Tg-135 K. These are considerably lower than the more traditional Tg-50 K values. This suggests that some relaxation processes are still active at relatively low temperatures. The kinetic fragility parameters (D*) were within the expected ranges.34,35 For dry ambient conditions, from the VFT model, the mean relaxation time for the free base at Tg−37 K is estimated to be about 460 h (19 days). Under the same dry conditions, the dihydrochloride salt at Tg-90 K is predicted to have an infinite mean relaxation time. Thus, based on the MDSC data, the dihydrochloride salt form seems to show a nontypical dual relaxation process. The first relaxation was fast and occurred on warming from the quench conditions without the need for any thermal annealing. The F

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics compact (closed) conformer to minimize interaction energies. In an interacting local environment, such as a polar solvent (water), tenapanor exists in an elongated (open) conformer. Representative open and closed conformers were selected for cFB and a-Di HCl, for full quantum mechanical structure minimization. The minimization was performed using density functional theory (DFT) with the basis set EDF2 6-31G*. The resulting minimized molecular structures are presented in Figures 7−10.

Figure 10. Minimized structure for closed tenapanor free base (DFT EDF2 6-31G*).

tenapanor is the best molecular unit for amorphous tenapanor, as can be seen in Figure 11. Figure 7. Minimized structure for open tenapanor dihydrochloride (DFT EDF2 6-31G*).

Figure 11. Final processed total scattering signal for noncrystalline tenapanor free base and dihydrochloride salt.

The lattice functions derived from the PXRD data of a-FB and a-Di HCl are shown in Figure 12. The positions of the large peaks at about 1.5 Q indicate there is more local order in a-FB than in a-Di HCl. Different solid structural models were used to calculate lattice functions and compared to the lattice functions derived from the measured data. The best fit was attained using a para-crystalline lattice function. Para-crystalline materials have short and medium range ordering in their lattices (similar to mesophases), putting them between frozen liquids and crystalline solids.36 In

Figure 8. Minimized structure for closed tenapanor dihydrochloride (DFT EDF2 6-31G*).

Figure 9. Minimized structure for open tenapanor free base (DFT EDF2 6-31G*).

Modeling was carried out using the Debye−Menke method. The Debye eq (eq 1) assumes an atomic-level molecular model of the entire sample, whereas the lattice function formalism (eq 2) utilizes a single average molecular unit arranged throughout the sample according to a random lattice function. The Debye−Menke total scattering curves calculated using the minimized molecular conformers of a-FB and a-Di HCl (in both open and closed forms) showed that the open conformer of

Figure 12. Lattice functions derived from the PXRD data of a-FB and aDi HCl. G

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

structure (Figure 7), while the amorphous free base develops the potential for three-dimensional stacking due to ordering achieved up to the third nearest neighbor that could result in achieving the closed energy minimized structure (Figure 10). Storage Stability of a-Di-HCl. The physical stability of aDi-HCl has been evaluated under ICH long-term storage stability conditions, and data collected for at least three years has shown no change in the PXRD pattern. Additionally, aged lots when tested with solid-state 13C NMR spectroscopy showed no spectral changes or appearance of crystallinity.

Figure 13, the calculated lattice functions are displayed with respect to a para-crystalline, one-dimensional, random lattice function for a-FB and a-Di-HCl.



CONCLUSIONS A variety of complementary techniques such as thermal analysis, powder X-ray diffraction, and molecular modeling were used to assess the crystallization propensity of two noncrystalline systems containing a relatively large MW drug substance, tenapanor (RDX5791). Tenapanor, is known to exist as crystalline anhydrous free base form (c-FB) and as an amorphous powder of dihydrochloride salt (a- Di-HCl). Molecular modeling and total scattering measurements suggested that the noncrystalline tenapanor dihydrochloride exists in an open elongated state with local 1D stacking in agreement with the para-crystalline model. The extent of stacking repeat is limited to its nearest-neighbor. Further, modulated DSC measurements were used to determine the relaxation dynamics of the noncrystalline forms below the glass transition event. The overall relaxation behavior was well described by the VFT stretched exponential model. The predicted mean relaxation time for the dihydrochloride salt under dry ambient conditions was infinite due to the observed high Tg. For tenapanor dihydrochloride, there is no evidence that a stable, crystalline form of lower energy than the relaxed glass can be formed. This is consistent with the observed physical stability data on several lots of tenapanor dihydrochloride under ICH accelerated and long-term storage conditions.

Figure 13. Derived lattice function for melt-quenched, noncrystalline tenapanor free base, and dihydrochloride plotted with respect to paracrystalline model.

The para-crystalline lattice function allowed two variables (stacking distance and damping factor) to be changed to alter the lattice function curve to better fit the measurement-derived curves. Each best-fitting model has a one-dimensional stacking distance of about 3.5 Å. Agreements between the calculated lattice functions and the para-crystalline models are reasonable, suggesting that some local nearest-neighbor alignment of the open tenapanor molecules is taking place in both a-FB and a-Di HCl. The para-crystalline stacking distance suggests that the molecules are lined up with respect to each other, effectively stacking normal to their long axes. The degree of local order is minimal and can be considered only a nearest-neighbor phenomena. PDF lattice functions were generated from the para-crystalline lattice functions, which provide a visual indication of the distances over which the molecules are ordering in the glassy state; (Figure 14). For the dihydrochloride, the local order



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.molpharmaceut.8b00853.



Additional experimental details on X-ray diffraction and solid state NMR data (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: 5104567717. ORCID

Radha R. Vippagunta: 0000-0003-2147-577X Author Contributions Figure 14. Lattice function PDF derived from noncrystalline free base and dihydrochloride lattice functions.

The manuscript was written through equal contributions of both authors. Both authors have given approval to the final version of the manuscript.

barely extends beyond nearest neighbor. The free base exhibits a more extensive range of local order out to at least the third nearest neighbor. The PDF lattice function calculations (Figure 14) can be used to conclude that the amorphous dihydrochloride salt only shows the potential for one-dimensional stacking that accounts for its open elongated energy minimized

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank Triclinic Laboratories for generating the data and their contributions. H

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics



(21) Angell, C. A. The old problems of glass and the glass transition, and the many new twists. Proc. Natl. Acad. Sci. U. S. A. 1995, 92, 6675− 6682. (22) Angell, C. A. Formation of Glasses from Liquids and Biopolymers. Science 1995, 267, 1924−1935. (23) Shamblin, S. L.; Chang, L.; Hancock, B. C.; Pikal, M. J. Characterization of the Time Scales of Molecular Motion in Pharmaceutically Important Glasses. J.Phys.Chem.B 1999, 103, 1413− 4121. (24) Bates, S.; Zografi, G.; Engers, D.; Morris, K.; Crowley, K.; Newman, A. Analysis of amorphous and nanocrystalline solids from their X-ray diffraction patterns. Pharm. Res. 2006, 23, 2333−2349. (25) Bhugra, C.; Pikal, M. J. Role of thermodynamic, molecular, and kinetic factors in crystallization from the amorphous state. J. Pharm. Sci. 2008, 97, 1329−1349. (26) Yoshioka, S.; Aso, Y. Correlations between molecular mobility and chemical stability during storage of amorphous pharmaceuticals. J. Pharm. Sci. 2007, 96, 960−981. (27) Ahlneck, C.; Zografi, G. The molecular basis for moisture effects on the physical and chemical stability of drugs in the solid state. Int. J. Pharm. 1990, 62, 87−95. (28) Alie, J.; et al. Dielectric study of the molecular mobility and the isothermal crystallization kinetics of an amorphous pharmaceutical drug substance. J. Pharm. Sci. 2004, 93, 218−233. (29) Feth, P. F.; Volz, J.; Hess, U.; Sturm, E.; Hummel, R. F. Physicochemical, crystallographic, thermal, and spectroscopic behaviour of crystalline and x-ray amorphous ciclesonide. J. Pharm. Sci. 2008, 97, 3765−3780. (30) Newman, A.; Engers, D.; Bates, S.; Ivanisevic, I.; Kelly, R. C. C.; Zografi, G. Characterization of amorphous API: Polymer mixtyres using X-ray powder diffraction. J. Pharm. Sci. 2008, 97, 4840−4856. (31) Okumura, T.; Otsuka, M. Evaluation of microcrystallinity of a drug substance, indomethacin, in a pharmaceutical model tablet by chemometric FT-Raman spectroscopy. Pharm. Res. 2005, 22 (8), 1350−1357. (32) Ueno, T.; et al. A Simple Method for Quantitative Determination of Active Drug Polymorphs and Amorphous in Drug Products by Fourier Transform-Raman Spectroscopy. Yakugaku Zasshi 2005, 125, 807−814. (33) Nilsson Lill, S. O.; Widdifield, C. M.; Pettersen, A.; Ankarberg, A. S.; Lind-kvist, M.; Aldred, P.; Gracin, S.; Shankland, N.; Shankland, K.; Schantz, S.; Emsley, L. Elucidating an Amorphous Form Stabilization Mechanism of Tenapanor Hydrochloride: Crystal Structure Analysis using X-ray Diffraction, NMR Crystallography and Molecular Modelling. Mol. Pharmaceutics 2018, 15 (4), 1476−1487. (34) Hancock, B. C.; Shamblin, S. L. Molecular mobility of amorphous pharmaceutical determined using differential scanning calorimetry. Thermochim. Acta 2001, 380, 95−107. (35) Zhou, D.; Grant, D. J. W.; Zhang, G. G. Z.; Law, D.; Schmitt, E. A. A Calorimetric Investigation of Thermodynamic and Molecular Mobility contributions to the Physical stability of Two Pharmaceutical Glasses. J. Pharm. Sci. 2007, 96, 71−83. (36) Atassi, F.; Mao, C.; Masadeh, A. S.; Byrn, S. R. Solid-state Characterization of Amorphous and Mesomorphous Calcium Ketoprofen. J. Pharm. Sci. 2010, 99, 3684−3697. (37) Scardi, P.; Billinge, S. J. L.; Neder, R.; Cervellino, A. Celebrating 100 years of the Debye scattering equation. Acta Crystallogr., Sect. A: Found. Adv. 2016, A72, 589−590. (38) Debye, P.; Menke, H. Phys. Z. 1930, 31, 797. (39) Graeser, K. A.; Patterson, J. E.; Rades, T. Applying thermodynamic and Kinetic Parameters to predict the physical stability of two differently prepared amorphous forms of Simvastatin. Curr. Drug Delivery 2009, 6, 374−382.

ABBREVIATIONS DFT, density functional theory; PXRD, powder X-ray diffraction; PDF, pair-wise distribution function



REFERENCES

(1) Babu, N. J.; Nangia, A. Solubility Advantage of Amorphous Drugs and Pharmaceutical Cocrystals. Cryst. Growth Des. 2011, 11, 2662− 2679. (2) Hancock, B. C.; Parks, M. What is the true solubility advantage for amorphous pharmaceuticals. Pharm. Res. 2000, 17, 397−404. (3) Hancock, B. C.; Zografi, G. Characteristics and significance of the amorphous state in pharmaceutical systems. J. Pharm. Sci. 1997, 86, 1− 12. (4) Yu, L. Amorphous pharmaceutical solids: Preparation, characterization and stabilization. Adv. Drug Delivery Rev. 2001, 48, 27−42. (5) Hancock, B. C.; Shamblin, S. L.; Zografi, G. Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures. Pharm. Res. 1995, 12, 799−806. (6) Zhou, D.; Zhang, G. G. Z.; Law, D.; Grant, D. J. W.; Schmitt, E. A. Physical stability of amorphous pharmaceuticals: Importance of configurational thermodynamic quantities and molecular mobility. J. Pharm. Sci. 2002, 91, 1863−1872. (7) Zhang, G. G. Z.; Zhou, D. Crystalline and Amorphous Solids-In Developing Solid Oral Dosage Forms, Pharmaceutical Theory & Practice. In Developing Solid Oral Dosage Forms, Pharmaceutical Theory & Practice; Qiu, Y., Ed.; Academic Press: London, 2017; p 49. (8) Hancock, B. C.; et al. Comparison of the mechanical properties of the crystalline and amorphous forms of a drug substance. Int. J. Pharm. 2002, 241, 73−85. (9) Almeida e Sousa, L.; et al. Assessment of the Amorphous “Solubility” of a Group of Diverse Drugs Using New Experimental and Theoretical Approaches. Mol. Pharmaceutics 2015, 12, 484−495. (10) Serajuddin, A. T. M. Solid dispersion of poor water-soluble drugs:Early promises, subsequent problems, and recent breakthroughs. J. Pharm. Sci. 1999, 88, 1058−1066. (11) Leuner, C.; Dressman, J. Improving drug solubility for oral delivery using solid dispersions. Eur. J. Pharm. Biopharm. 2000, 50, 47− 60. (12) Kothari, K.; Ragoonanan, V.; Suryanarayanan, R. Influence of Molecular Mobility on the Physical Stability of Amorphous Pharmaceuticals in the Supercooled and Glassy States. Mol. Pharmaceutics 2014, 11, 3048−3055. (13) Gupta, P.; Chawla, G.; Bansal, A. K. Physical stability and solubility advantage from amorphous celecoxib: The role of thermodynamic quantities and molecular mobility. Mol. Pharmaceutics 2004, 1, 406−413. (14) Johari; Goldstein. Viscous liquids and the glass transition.II. Secondary relaxations in glasses of rigid molecules. J. Chem. Phys. 1970, 53, 2372. (15) Mehta, M.; et al. Effect of Water on Molecular Mobility and Physical Stability of Amorphous Pharmaceuticals. Mol. Pharmaceutics 2016, 13, 1339−1346. (16) Munch, E.; Pelletier, J.-M.; Vigier, G. Increase in molecular mobility of an amorphous polymer deformed below Tg. J. Polym. Sci., Part B: Polym. Phys. 2008, 46, 497−505. (17) Zhou, D.; Zhang, G. G. Z.; Law, D.; Grant, D. J. W.; Schmitt, E. A. Thermodynamics, Molecular Mobility and Crystallization Kinetics of Amorphous Griseofulvin. Mol. Pharmaceutics 2008, 5 (6), 927−936. (18) Zhou, D.; et al. Physical stability of amorphous pharmaceuticals: Importance of configurational thermodynamic quantities and molecular mobility. J. Pharm. Sci. 2002, 91, 1863−1872. (19) Baird, J. A.; Van Eerdenbrugh, B.; Taylor, L. S. A classification system to assess the crystallization tendency of organic molecules from undercooled melts. J. Pharm. Sci. 2010, 99, 3787−3806. (20) Van Eerdenbrugh, B.; Baird, J. A.; Taylor, L. S. Crystallization tendency of active pharmaceutical ingredients following rapid solvent evaporation–classification and comparison with crystallization tendency from undercooled melts. J. Pharm. Sci. 2010, 99, 3826−3838. I

DOI: 10.1021/acs.molpharmaceut.8b00853 Mol. Pharmaceutics XXXX, XXX, XXX−XXX