2024 Volume 72 Issue 12 Pages 1073-1083
A co-amorphous state composed of probucol (PC) and fluvastatin sodium salt (FLU) was prepared by spray-drying (SD). We have previously reported that PC and atorvastatin calcium trihydrate salt (ATO) formed a co-amorphous state when prepared by a SD method and that the solubility of PC and the amorphous stability were improved by the preparation of the co-amorphous state. In the present study, the physicochemical properties, including the amorphous stability of the prepared co-amorphous state, were characterized. Powder X-ray diffraction measurement results suggested that PC and FLU formed a co-amorphous state and that a higher percentage of PC was dissolved from the PC–FLU co-amorphous state than from the PC–ATO co-amorphous state. The results are attributed to FLU exhibiting greater solubility and wettability than ATO, which is supported by the results of solubility tests and contact-angle measurements. The stability of the amorphous state of PC is higher in the co-amorphous state with ATO than in that with FLU. This difference is attributed to differences in the molecular interaction modes between PC–FLU and PC–ATO. Therefore, the selection of high-wettability molecules as a co-former for the co-amorphous state enhances its water solubility. The present study also indicates that molecular interactions enhance the stability of the co-amorphous state.
Most of the recently developed active pharmaceutical ingredients (API) and drug candidate compounds developed through combinatorial chemistry and high-throughput screening are poorly soluble in gastrointestinal tract fluid.1,2) In addition, up to 40% of the drugs already on the market are poorly soluble in water and exhibit low bioavailability.3) Most of these drugs are classified as Biopharmaceutical Classification System (BCS) class II drugs,4) which have good cell membrane permeability but poor water solubility.3,5,6) The therapeutic efficacy of solid dosage forms such as tablets and capsules is dependent on the bioavailability of the drug, which is, in turn, affected by the solubility and dissolution rate of the drug at the site of absorption.7) Therefore, several strategies for improving the solubility of BCS class II drugs, including the preparation of drugs as solid dispersions,8) inclusion complexes of drugs with cyclodextrin,9) and the formation of co-crystals10) and co-amorphous compounds,11) have been investigated to enhance the bioavailability of drugs.5,12)
Probucol (PC), an anticholesterol drug, is a typical BCS class II compound with low water solubility (5 ng/mL) and limited oral bioavailability (<10%).5,13) Therefore, effective strategies for improving the solubility of PC are needed and several methods have been investigated.5,14,15) For example, Yagi et al. and Kubo et al. reported that for a solid dispersion of PC and the hydrophilic polymer polyvinylpyrrolidone (PVP), a significant increase in both the solubility and bioavailability of PC was observed.16,17) The optimum PC-to-PVP weight ratio was found to be 1 : 9. As mentioned above, solid dispersions are useful for improving drug solubility and maintaining the stability of the amorphous phase. However, there are issues regarding the large tablet or capsule size for these formulations due to the larger amount of carrier required. Therefore, we have focused on a co-amorphous formulation that contain a poorly soluble drug and a low-molecular-weight compound to improve solubility and stability. The improvement in drug solubility is due to interactions between these two components.18) In addition, the amorphous state of a co-amorphous formulation is expected to remain stable for longer than that for a single amorphous component due to restricted molecular motion.18) Due to the improved drug solubility, the size of tablets and capsules can also be reduced compared to the case where an additive polymer is used.18) We have previously reported that the solubility of PC is improved by forming a co-amorphous state with atorvastatin calcium salt trihydrate (ATO) via a spray-drying (SD) method.18) We found that ATO and PC underwent molecular interactions such as hydrogen bonding in the co-amorphous state and that the PC exhibited greater solubility than raw PC crystals.18) In addition, the co-amorphous phase could be stored for 35 d without crystallization occurring, even under stress conditions of 40 °C and 75% relative humidity (RH). In the co-amorphous phase, the PC-to-ATO weight ratio was about 1.0 : 2.3, thus allowing the size of tablets or capsules to be reduced compared to that for solid dispersions containing a polymer.
In the present study, we focused on fluvastatin sodium salt (FLU), which like ATO is categorized as a statin group, and attempted to prepare a co-amorphous phase composed of PC and FLU. FLU has been used to treat hypercholesterolemia by lowering the concentration of low-density lipoprotein (LDL), cholesterol, serum triglycerides, and apolipoprotein B,19–21) and is categorized as BCS class II because of its poor solubility in water (3.9 µg/mL).19,22–24) In the present study, we used FLU as a co-former, instead of the free form of fluvastatin. Given the fact that FLU is used in medicinal formulations, the ability to prepare it in co-amorphous form is expected to be useful for clinical applications. In addition, since the charge state of the co-former is expected to have a significant effect on the state of the co-amorphous phase, we compared the properties of co-amorphous PC–FLU with those of the previously reported PC–ATO.18) We attempted to prepare a co-amorphous phase of PC and FLU that could improve the solubility of both PC and FLU, which exhibit low bioavailability because of their poor solubility in water.19,25)
Sasaki et al. reported the effects of a combination treatment of PC and FLU administered as individual dosage forms.26) They investigated the efficacy and safety of FLU in hypercholesterolemic patients already receiving PC and reported that the addition of FLU to the PC regimen further significantly reduced serum total and LDL cholesterol concentrations. These effects were fully established within 4 weeks after treatment and were maintained throughout the treatment period. The authors evaluated 27 patients who participated in the study, all of whom were evaluated for safety and 22 of whom were evaluated for efficacy. They concluded that a combination of PC and FLU can be considered an effective and well-tolerated treatment for hypercholesterolemia. Given this information, FLU is expected to be used in combination with PC treatment in the future.
We prepared amorphous PC with FLU using the SD method described in our previous report and evaluated the crystallinity of the samples by powder X-ray diffraction (PXRD) and differential scanning calorimetry (DSC).18) The particle surface morphology and shape were observed by scanning electron microscopy (SEM), and molecular interactions in samples were investigated by Fourier transform (FT) IR spectroscopy. We also conducted dissolution and storage stability tests to evaluate the solubility and physicochemical stability of PC in the prepared samples, respectively. In addition, the solubility of FLU in the dissolution medium and the contact angle for samples were measured to understand the results of the PC dissolution tests. One of the major goals of this study is to compare the usability of co-amorphous PC–FLU and PC–ATO by analyzing their physicochemical properties, including their dissolution behavior after being stored in a stress environment.
PC [MW 516.84, Fig. 1(a)], FLU [MW 433.46, Fig. 1(b)], and ATO [MW 1209.41, Fig. 1(c)] were purchased from Tokyo Chemical Industry (Tokyo, Japan). Acetonitrile (HPLC grade) was purchased from Nacalai Tesque (Kyoto, Japan). Ethanol (reagent grade) and polyoxyethylene sorbitan monooleate (Tween 80) were purchased from Kishida Chemical (Osaka, Japan). All other chemicals used were of reagent grade.
The formulations of physical-mixture (PM) and SD samples are shown in Table 1. The PC-to-FLU molar ratio in the PM and SD samples was 1 : 2, whereas the PC-to-ATO ratio was 1 : 1, as previously reported.18) These ratios differ because ATO is formed by two molecules of free-form atorvastatin and one calcium ion [Fig. 1(c)], whereas FLU is formed by one free-form fluvastatin molecule and one sodium ion [Fig. 1(b)]. Therefore, the molar ratios of PC and the free form of the drugs (fluvastatin or atorvastatin) are the same at 1 : 2 (PC: fluvastatin or atorvastatin free form). One of the objectives of the present study is to compare the properties of co-amorphous PC–FLU with those of co-amorphous PC–ATO reported previously. Therefore, we considered it necessary to use the same number of co-former (ATO and FLU) and PC molecules, and set the PC-to-FLU molar ratio to 1 : 2 in the present study. Considering the possibility that FLU and ATO may interact with PC in a significantly different manner, we also prepared a sample with a PC-to-FLU molar ratio of 1 : 1 and examined the formation of the co-amorphous phase. The results of preliminary tests (Supplementary Figs. S1, S2) indicated that it was difficult to form a co-amorphous phase with a PC-to-FLU molar ratio of 1 : 1. Therefore, we set the ratio to 1 : 2 in subsequent experiments. The PM samples were prepared by physically mixing PC and FLU using a mortar and pestle. Amorphous PC was prepared by the same method as in a previous study.18) Briefly, PC was dissolved in ethanol, and the solvent was then evaporated at 100 °C, after which the residue was heated at 150 °C using a sample concentrator (BSR-MD200-1, Bio Medical Science Co., Ltd., Tokyo, Japan). The residue was rapidly cooled to −20 °C and collected as amorphous PC. Amorphous FLU was prepared by SD as follows27): 1 g of FLU was dissolved in 20 mL of ultrapure water, the solution was filtered using a 0.2 µm membrane filter, and SD was performed using a spray dryer (GS-310, Yamato Scientific, Tokyo, Japan) to produce amorphous FLU. The inlet temperature of the drying chamber was maintained at 140 °C. The amorphous PM sample was prepared by physically mixing amorphous PC and amorphous FLU using a mortar and pestle at a PC-to-FLU molar ratio of 1 : 2. As previously reported,18) SD samples were prepared using a spray dryer by the following procedure. PC and FLU or ATO were dissolved in ethanol solution and the drug solution was sprayed into the chamber of the spray dryer through a nozzle with a diameter of 400 µm at a rate of 10 mL/min and a spray pressure of 0.15–0.2 MPa. The inlet temperature of the drying chamber was maintained at 50 °C. A co-amorphous phase composed of PC and ATO (SD_PC_ATO) and a sample of spray-dried PC (SD_PC) for use in the stability tests (see “Stability Tests”) were also prepared according to previously reported methods.18) The PC-to-ATO molar ratio in SD_PC_ATO was 1 : 1; SD_PC was obtained by dissolving PC in ethanol and spray-drying it.18)
| Sample name | Mole ratio | PC (g) | FLU (g) | Ethanol (mL) |
|---|---|---|---|---|
| PM_PC_FLU | PC : FLU = 1 : 2 | 1.0000 | 1.6773 | 0 |
| SD_PC_FLU | PC : FLU = 1 : 2 | 1.0000 | 1.6773 | 268 |
PXRD measurements were performed on an Ultima IV X-ray diffractometer (Rigaku, Tokyo, Japan). The measurement conditions were as follows: graphite-monochromated CuKα radiation (λ = 1.54178 Å); 40 kV and 40 mA; scanning interval of 3–40° (2θ); and scanning speed of 2°/min. The PDXL 2.9 (Rigaku, Tokyo, Japan) software was used to analyze the PXRD patterns.
Thermal AnalysisDSC was performed on samples (5.0 ± 0.05 mg) in open aluminum pans using a thermal analyzer system (DSC-60 Plus, Shimadzu, Kyoto, Japan) with Al2O3 used as a reference. In this study, we used indium metal to calibrate the temperature and heat flow in the device. The samples were heated from 25 to 200 °C at 10 °C/min under flowing N2 gas (50 mL/min). The melting point was determined as the onset temperature for the peak. To measure the glass transition temperature (Tg), corresponding samples were measured from −10 to 215 °C at 10 °C/min using crimped aluminum pans, and Tg was determined as the midpoint of the heat capacity increase. LabSolutions TA version 1.01 (Shimadzu) software was used to analyze the DSC thermograms.
Morphology ObservationsThe particle surface morphology and shape were examined by SEM. The samples were braced on carbon tape and then sputtered with a Pt layer in a cathodic evaporator operated at a current of 35 mA for 1.5 min. Images were obtained using a JXA-iHP200F field-emission electron probe microanalyzer (Jeol, Tokyo, Japan).
FTIR SpectroscopyFTIR spectra of samples were obtained in the wavenumber range from 400 to 4000 cm−1 using an FTIR spectrometer (FTIR-8400S, Shimadzu). The FTIR measurements were conducted via the KBr method, where samples for measurement were mixed with KBr and the resultant mixtures were compacted into pellets. Spectra were collected using 32 scans per spectrum at a resolution of 4 cm−1.
Dissolution TestsDissolution tests were conducted as described in a previous study18) and in accordance with the Japanese Pharmacopoeia (JP) X VIII using an NTR-3000 dissolution tester (Toyama Sangyo, Osaka, Japan). The dissolution medium was prepared by dissolving Tween 80 in distilled water to a concentration of 1% (w/v%). For PC dissolution tests, samples containing 25 mg of PC (theoretically calculated on the basis of the formulations in Table 1) were added to 900 mL of dissolution medium with a paddle and the resultant mixtures were stirred at a rotation speed of 100 rpm at 37 °C, as specified in JP X VIII. For ATO dissolution tests, the sample (theoretically containing 58.5 mg of ATO) was added to the dissolution medium and the dissolution test was performed under the same conditions as those used for PC. These dissolution tests were conducted under the same conditions as those previously reported.18) For FLU, the sample (theoretically containing 41.9 mg of FLU) was added to the dissolution medium and the dissolution test was performed under the same conditions used for PC and ATO. At predetermined time intervals, 5 mL samples were withdrawn and replaced with fresh dissolution medium. Similar to previous studies, dissolution tests of PC were performed under non-sink conditions and dissolution tests of ATO were performed under sink conditions. For FLU, dissolution tests were conducted under sink conditions equal to approximately 0.242% (w/v) of the saturated solubility of raw FLU crystals. The collected samples were filtered (pore size, 0.45 µm) and then diluted using ethanol. The concentrations of PC, FLU, and ATO were measured by HPLC. Dissolution tests were conducted for samples on the day of preparation and after the samples were stored for a defined period under stability test conditions (see “Stability Tests”).
HPLC ConditionsThe HPLC system comprised a Prominence LC-20AD intelligent HPLC pump, an SPD-20A intelligent UV/VIS detector, a CTO-10AS intelligent column oven, an SIL-10AF intelligent sampler, a CBM-20A system controller, and an LC solution chromatography data system (all from Shimadzu). The HPLC conditions for PC concentration determination were as follows: column, YMC-Pack Pro C18 4.6 × 150 mm2 (YMC, Kyoto, Japan); temperature, 40 °C; detector, UV; wavelength, 242 nm; flow rate, 1.0 mL/min; injection volume, 20 µL; and mobile phase, acetonitrile:ultrapure water = 93 : 7. The HPLC conditions for the determination of PC, FLU, and ATO concentrations using an HPLC linear gradient program were as follows: column, YMC-Pack Pro C18 4.6 × 150 mm2 (YMC); temperature, 40 °C; detector, UV; wavelength, 244 nm; flow rate, 1.0 mL/min; injection volume, 20 µL; and mobile phase, gradient conditions of mobile phases A and B. Mobile phase A was prepared by dissolving citric acid monohydrate in ultrapure water to a concentration of 1.05% (w/v%, pH: 4.0 adjusted with ammonia water). Mobile phase B was acetonitrile. The HPLC gradient program (time (min)/v/v% mobile phase B) is shown in Supplementary Table S1. As described above, two sets of conditions were used for measurement of the concentration of PC, and the measurements were confirmed to have been conducted with the same accuracy.
Solubility TestsSolubility tests for FLU were carried out by mechanical shaking using a thermo-shaker (BSR-MSC100, Bio Medical Science Co., Ltd., Tokyo, Japan). An excess amount of crystalline drug powder (FLU: 100 mg) was added to 5 mL of the dissolution medium used in the dissolution tests, and the resultant mixture was mechanically shaken. The shaking conditions were as follows: temperature, 37 °C; shaking time, 48 h; and revolutions, 1000 rpm. After shaking, the samples were filtered (pore size, 0.45 µm) and suitably diluted with ethanol; the concentration of the samples was then analyzed by HPLC.
Contact-Angle MeasurementsContact angles were measured using an image analysis method.28–30) A PC or SD sample (10 mg) was added to 1 mL of ethanol and dissolved. To form a drug layer on the cover glass, 100 µL of the ethanolic drug solution was dropped onto the cover glass and dried at room temperature. The films prepared on cover glasses using raw PC crystals, SD_PC_ATO, and SD_PC_FLU were named PC film, PC_ATO film and PC_FLU film, respectively. Twenty microliters of dissolution medium or water were dropped from a height of approximately 1 cm onto the drug layer formed on the cover glass and photographed from the side using a dual 12MP camera system, and wide-angle (seven-element lens) and ultra-wide angle (five-element lens) cameras (iPhone 12 mini, Apple Inc.). The contact angle was measured from the images using the ImageJ image analysis software.31)
Stability TestsStability tests were conducted under accelerated conditions. The tests were performed for samples stored in a stress environment at 40 °C and 75% RH (saturated solution of NaCl). The crystallinity of the samples was assessed by PXRD and DSC analyses after a defined period of storage. Dissolution tests and morphology observations were also conducted after a defined period of storage.
PXRD patterns for raw FLU crystals, PM_PC_FLU, amorphous FLU, amorphous PM and SD_PC_FLU are shown in Fig. 2. PC has been reported to have several crystal forms.32,33) The crystal form for raw PC crystals and SD_PC is form I and form II, respectively.18) Figure 2 shows that PM_PC_FLU is crystalline because its PXRD pattern includes peaks derived from PC and FLU [Fig. 2(b)]. Unfortunately, the PXRD pattern for amorphous FLU shows a peak at a 2θ value of around 3° [Fig. 2(c)], even though the sample was prepared with reference to a patent for the preparation of amorphous FLU published by Bhanu et al.27) Another patent for producing amorphous FLU reported a PXRD pattern that was similar to that obtained in the present study.34) We therefore concluded that the sample prepared in the present study was amorphous FLU. The sample produced by PM showed a PXRD pattern associated with both amorphous PC and amorphous FLU [Fig. 2(d)]. On the other hand, the PXRD pattern for SD_PC_FLU shows a halo pattern with no peaks derived from PC or FLU [Fig. 2(e)]. It was found that an SD sample (SD_PC_FLU_1_1) with a PC-to-FLU molar ratio of 1 : 1, which was prepared for the preliminary tests, also showed a halo pattern similar to that for SD_PC_FLU [Supplementary Fig. S1]. The PXRD results for SD_PC_FLU and SD_PC_FLU_1_1 in the present study indicate that PC can feasibly form a co-amorphous phase with FLU via SD.
DSC thermograms for raw FLU crystals and PM_PC_FLU are shown in Fig. 3(I). The onset melting points for PC forms I and II have been reported to be 126 and 116 °C, respectively.33) We have confirmed in a previous study that the onset melting points for raw PC crystals and SD_PC are 125.88 and 115.63 °C, respectively.18) Regarding the DSC thermogram for FLU, various temperatures such as 146.88 or 195 °C have been reported as the melting point for FLU.35,36) In our study, no apparent melting point for FLU was detected in the temperature range from 25 to 200 °C when using open aluminum pans [Fig. 3(I) (a)]. On the other hand, the onset melting point for raw FLU crystals was detected at 190.52 ± 5.67 °C when using crimped aluminum pans (n = 9). An example of a DSC thermogram for raw FLU crystals measured with crimped aluminum pans is shown in Supplementary Fig. S3. The differences among the reported FLU melting points are likely attributable to differences in the measurement conditions and crystal structures of the samples. In the thermogram for the PM_PC_FLU sample, an endothermic peak was detected at 125.89 °C; this temperature approximately coincides with the melting point for raw PC crystals [Fig. 3(I) (b)]. Therefore, we considered that the PC in the PM_PC_FLU sample did not undergo a polymorphic transformation and remained in form I. The amorphous FLU showed a gentle exothermic peak between 170 and 200 °C, an endothermic peak at 203.14 °C [Fig. 3(II) (c)], and a Tg of 58.24 °C [Fig. 3(III) (c)]. Amorphous PM also showed an exothermic peak due to recrystallization at around 58.41 °C, in addition to an endothermic peak at 116.18 °C that is considered to be associated with melting of PC form II [Fig. 3(II) (d)]. For the amorphous PM sample, Tg was determined to be 27.49 °C [Fig. 3(III) (d)]. We previously reported that PC in amorphous form has a Tg of 28.41 °C, when measured with a heating rate of 10 °C/min.18) Therefore, the Tg value determined for amorphous PM could be attributed to the glass transition of PC. In the amorphous PM sample, a peak associated with Tg for FLU was not detected because it overlapped with the exothermic peak due to the recrystallization of PC at 58.41 °C. The SD_PC_FLU sample showed an exothermic peak at 55.77 °C, which is considered to be associated with recrystallization of PC, and very small endothermic peaks at 113.45 and 122.98 °C, which are considered to be derived from PC form II and form I, respectively, and it is considered that the reduction in melting point is due to the coexistence of FLU [Fig. 3(II) (e)]. The melting enthalpy values for PC in SD_PC_FLU were 1.66 and 0.37 J/g, whereas that for PM_PC_FLU was 21.85 J/g, which indicates that the crystals in SD_PC_FLU are much smaller than those in PM_PC_FLU. In addition, a halo PXRD pattern was observed for SD_PC_FLU. Therefore, we consider that a small amount of PC forms I and II existed in the almost amorphous SD_PC_FLU. On the other hand, in the preliminary tests, the DSC thermogram for SD_PC_FLU_1_1 exhibited a larger endothermic peak than that for SD_PC_FLU [Supplementary Fig. S2], and was more similar to that for amorphous PM than that for SD_PC_FLU. Therefore, it is considered that the SD_PC_FLU_1_1 sample contained a single-phase amorphous component, whereas the SD_PC_FLU sample consisted almost entirely of a co-amorphous phase. Regarding the glass transition temperature, SD_PC_FLU exhibits much higher Tg values that are considered to be derived from PC and FLU (48.80 and 89.41 °C, respectively) than amorphous PC (28.41 °C) or amorphous FLU (58.24 °C) [Fig. 3(III) (e)]. Therefore, the amorphous states of PC and FLU in SD_PC_FLU are expected to be more stable than those in each amorphous PC or FLU. We have previously reported that a co-amorphous phase is formed by PC and ATO and that no endothermic peak appears in the DSC thermogram for SD_PC_ATO.18) For this co-amorphous phase, transition points attributable to Tg were detected at 48.34, 81.98, 122.92, and 162.98 °C, and the Tg value for PC in the co-amorphous sample was 48.34 °C. As described above, the Tg for PC in SD_PC_FLU in the present study was considered to be 48.80 °C.18) Comparing the Tg values for the two co-amorphous materials, there is no significant difference in the Tg value for PC, but the number of transitions detected is different. We speculate that molecular interaction mechanisms in SD_PC_FLU might differ from those in SD_PC_ATO, leading to the observed difference in crystallinity and observed Tg. In their review paper, Dengale et al. reported several cases of co-amorphous formulations that have higher Tg values than those for the respective individual components.37) They also indicated that the increase in physical stability cannot be explained by Tg alone, but must involve other factors, including molecular interactions such as hydrogen bonding and/or π–π interactions between the components in the co-amorphous mixture, and molecular-level mixing. Löbmann et al. reported that a co-amorphous phase formed by simvastatin (SVS) and glipizide (GPZ) showed improved stability that was attributable to molecular-level mixing during milling, with GPZ acting as an anti-plasticizer in these mixtures.38) They also indicated that samples with only a single Tg value were well-mixed binary systems containing only one amorphous phase, whereas samples with two Tg values were systems with multiple amorphous phases, each having a different SVS-to-GPZ ratio. Laitinen et al. reported that there was no clear evidence of molecular interactions in a co-amorphous mixture formed by SVS and lysine (LYS).39) They attributed the increased stability of the co-amorphous sample to molecular mixing of SVS and LYS, and observed a higher Tg for the co-amorphous mixture than that for amorphous SVS. Based on the above literature reports, it is believed that the increase in Tg may not only be due to molecular interactions, but also to mixing at the molecular level, and the miscibility of the drug and the co-former may also have a significant effect on Tg and the stability of the samples. In the present study also, the two Tg values for the SD_PC_FLU sample could be due to the formation of two amorphous regions with different PC-to-FLU ratios. In both the present and previous studies, we prepared co-amorphous samples by SD. However, it is difficult to determine the mechanisms involved in the formation of the co-amorphous phase and phase separation during the SD process. Phase separation might be the result of different solubilities and drying rates for the different components (PC, ATO, and FLU) in ethanolic droplets during the preparation of SD particles. This would lead to several Tg values associated with amorphous regions with different PC-to-co-former ratios.
SEM images of the morphology of the raw PC crystals, raw ATO crystals, raw FLU crystals, SD_PC, SD_PC_ATO, and SD_PC_FLU are shown in Fig. 4. The raw PC, raw ATO, and raw FLU contain many crystals that are larger than 10 µm and appear to have irregular shapes [Figs. 4(a), (b), and (c)]. In addition, the raw ATO and raw FLU contain numerous flaky crystals [Figs. 4(b), 4(c)]. By contrast, the particle size for the SD samples appears to be smaller than that for the raw PC crystals and co-formers [Figs. 4(d)–4(f)]. The SEM image of SD_PC shows particles of various sizes and shapes aggregated together, whereas the SEM image of the SD_PC_ATO shows nonspherical particles with dents [Figs. 4(d), 4(e)]. By contrast, the particles of SD_PC_FLU exhibit a spherical shape with a smooth surface [Fig. 4(f)]. We have previously reported that SD particles composed of indomethacin and an amphiphilic polymer, Soluplus®, differed in shape depending on the solution composition during the SD process, and that some of the particles were nonspherical with deep cavities.8) Different particle morphologies are considered to be caused by differences in the volatility of the solvents used in the synthesis of the samples and differences in the solid density of the droplets.8) In the present study, the SD particles were all prepared using ethanol as the solvent. Therefore, the different particle morphologies were likely caused by the difference in solid density or viscosity of the SD solutions.
We have previously reported that new molecular interactions, such as hydrogen bonding, occur between PC and ATO in co-amorphous PC–ATO.18) The interactions between drug molecules are considered to affect the physicochemical properties and stability of the co-amorphous state. In the present study, we conducted FTIR measurements of SD_PC_FLU to investigate whether PC and FLU undergo molecular interactions in the co-amorphous state. FTIR spectra of raw FLU crystals, amorphous FLU, PM_PC_FLU, amorphous PM and SD_PC_FLU are shown in Supplementary Figs. S4(A) (4000–2400 cm−1) and S4(B) (1800–400 cm−1). As previously reported, the FTIR spectra of raw PC crystals show a characteristic O–H stretching peak at 3634 cm−1, that shifts to 3626 cm−1 for SD_PC along with the appearance of a new peak at 3545 cm−1 due to the formation of weak intermolecular O − H···S hydrogen bonds.18) The FTIR spectrum of raw FLU crystals has been reported to show several characteristic peaks, including peaks at 3400 cm−1 due to O–H stretching, 1582 cm−1 due to C=O stretching, 1155 cm−1 due to C–O stretching, 1215 cm−1 due to C–N, and 966 cm−1 due to aryl–F.35) The FTIR spectra of PM_PC_FLU and SD_PC_FLU (Supplementary Fig. S4) show only slight changes compared with the spectra of raw PC crystals and raw FLU crystals. As shown in Supplementary Fig. S4, a characteristic peak due to C–N was observed at 1215 cm−1 in the FTIR spectrum of raw FLU crystals, and corresponding peaks were observed at 1217, 1223, 1221, and 1219 cm−1 for amorphous FLU, PM_PC_FLU, amorphous PM, and SD_PC_FLU, respectively. It is considered that only weak molecular interactions involving C–N bonds occurred in FLU due to contact with PC molecules, and no strong interactions were associated with amorphization because no significant peak shift was detected in the amorphous FLU spectrum. It is therefore likely that no significant molecular interactions occurred between PC and FLU in SD_PC_FLU. Although it remains uncertain why different molecular interactions occurred between PC and the co-formers FLU and ATO, it is possible that this is related to the different chemical structures of the co-formers. The chemical structures of FLU and ATO are similar in many respects, but ATO contains an amide group whereas FLU does not [Figs. 1(b), 1(c)]. In a previous study, we found that molecular interactions such as hydrogen bonding occurred between PC and the amide groups of ATO in the SD_PC_ATO sample.18) Amide groups are known to be capable of forming hydrogen bonds, so that ATO is more likely than FLU to interact with PC. The difference in chemical structure between FLU and ATO may therefore have affected the interaction with PC. Löbmann et al. reported that a co-amorphous mixture of two BCS Class II drugs, SVS and GPZ, was obtained by ball milling and cryo-milling, and that no specific interactions occurred between SVS and GPZ molecules in the mixture.38) Therefore, in some cases, clear molecular interactions may not occur even in a co-amorphous state. The DSC and FTIR results suggest that PC–FLU and PC–ATO exist in different molecular configurations in SD_PC_FLU and SD_PC_ATO, respectively, even though FLU and ATO have similar molecular structures. In the DSC profile for SD_PC_FLU shown in Fig. 3(II), an exothermic peak associated with recrystallization of PC appears at 55.77 °C, together with a very weak endothermic peak due to melting. Because two Tg values and an exothermic peak were detected followed by a melting point in SD_PC_FLU, the sample may contain two kinds of co-amorphous phases, one rich in PC and the other rich in FLU. This is based on the assumption that during the DSC measurements, only PC recrystallized from the PC-rich co-amorphous phase and then melted. The PC-rich phase is considered to contain a large number of relatively free PC molecules with no molecular interactions.
Dissolution TestsThe release profiles for PC from PM_PC_FLU and SD_PC_FLU are shown in Fig. 5(I) and those for FLU from PM_PC_FLU and SD_PC_FLU are shown in Fig. 5(II). We have previously reported that the PC released from raw PC crystals and SD_PC was present at concentrations below the detection limit or was detected at low concentrations until 2 h after the start of the dissolution test.18) The results of dissolution tests of PM_PC_FLU show that the PC released from the sample is below the detection limit. However, PC released from SD_PC_FLU was detected at a concentration of 18.2 ± 1.2 µg/mL, and the percentage of dissolved PC was approx. 63% of the amount loaded (Fig. 5(I)). On the other hand, Fig. 5(II) shows that the FLU released from SD_PC_FLU reached about 100% of the amount loaded within 7 min, and that from PM_PC_FLU also reached over 90% within 7 min, which showed that for both samples, the FLU dissolution rate was higher than that for PC. However, as previously noted, a large difference was observed in the percentage of released PC between PM_PC_FLU and SD_PC_FLU. The SD_PC_FLU was prepared by the SD method, and the PC and FLU became molecularly mixed in the solution during the preparation process. Therefore, the molecular arrangement of PC and FLU in SD_PC_FLU was disordered, resulting in a halo PXRD pattern (Fig. 2(e)); these phases are considered to be in a nearly amorphous state. In our previous study, SD_PC_ATO, in which PC and ATO were mixed at the molecular level, showed a higher percentage of dissolved PC (approx. 9% of the amount loaded) than a physical mixture of PC and ATO.18) In the present study, the results of the dissolution tests showed that, like the previously reported PC released from SD_PC_ATO,18) that released from SD_PC_FLU is in a supersaturated state for at least 2 h. Interestingly, the amount of PC dissolved from SD_PC_FLU, which contains a small amount of PC form I and II crystals in a nearly co-amorphous state, was greater than the amount dissolved from SD_PC_ATO, which is in a co-amorphous state. This result suggests that, in addition to amorphization, other factors might play an important role in improving the solubility of PC in SD_PC_FLU. Jensen et al. reported the solubility and physicochemical properties of co-amorphous mixtures of indomethacin and amino acids using SD. They found that the dissolution process for the mixture is influenced by the strength of molecular interactions and the difference in solubility between the two components in the mixture.40) Therefore, we consider that the solubility of FLU in the dissolution medium affects the solubility of PC in SD_PC_FLU. To elucidate the mechanism by which the solubility of PC in SD_PC_FLU is improved, we conducted solubility tests for FLU in the dissolution medium and measured the contact angle for the dissolution medium and water on a film of PC, SD_PC_FLU, or SD_PC_ATO.
In a previous study, we conducted solubility tests and found that the solubility of raw PC crystals and raw ATO crystals in the same dissolution medium as that used in the present study was 25.6 ± 8.8 and 781.2 ± 13.8 µg/mL, respectively.18) In the present study, we found that the solubility of raw FLU crystals in the dissolution medium is 19.2 ± 0.4 mg/mL. PC, ATO, and FLU are all categorized as BCS class II.15,19,41) The solubility tests confirmed that the solubility of ATO and FLU differs substantially: the solubility of FLU in the dissolution medium is much higher than that of ATO even though ATO and FLU are both categorized as statin drugs. These results explain why a higher percent of PC was dissolved from SD_PC_FLU than from SD_PC_ATO. Villeda-Villegas et al. reported that a co-amorphous mixture consisting of pioglitazone hydrochloride and FLU could be obtained by ball milling, and investigated the physicochemical properties and intrinsic solubility of FLU.42) They found that the amount of FLU released from the co-amorphous mixture was low, and attributed this to the low solubility of pioglitazone hydrochloride. Dengale et al. reported on the mechanism of co-amorphous material formation and stabilization, and briefly reviewed other studies on the effect of co-amorphization on dissolution and in vivo performance and the future potential of co-amorphization.37) They concluded that the dissolution rate for a poorly soluble drug from a co-amorphous mixture with a co-former depends on the solubility of the co-former and the strength of the molecular interactions between the drug and the co-former. Therefore, we consider that the difference in solubility between ATO and FLU is one of the reasons why the dissolution rate for the PC in SD_PC_FLU is higher than that for the PC in SD_PC_ATO.
Contact AngleWe conducted contact-angle measurements of film samples to explore why the percentage of PC dissolved from SD_PC_FLU was higher than that from SD_PC_ATO (Fig. 5).18) Karavas et al. reported that the contact angle is a representative indicator of the surface polarity and hydrophilicity of a system; they used contact-angle measurements to investigate the effect of the hydrophilicity of PVP in a solid dispersion of PVP and felodipine.28) Lu et al. also noted that wettability is usually indicated by contact angle; they conducted contact-angle measurements to elucidate the relationship between wettability and dissolution in a solid dispersion of SVS and PVP.30) They reported that the contact angle decreased with increasing PVP content and that the dissolution efficiency of the solid dispersion increased with decreasing contact angle.30) They therefore concluded that wettability is indeed one of the most important factors influencing the dissolution of solid dispersions.30) Sóti et al. prepared solid dispersions of itraconazole and Eudragit E using three methods and conducted contact-angle measurements to investigate the differences in dissolution profiles for samples with different drug contents.43) They attributed slightly slower drug release rates to reduced wettability of the samples, as indicated by their higher contact angles.43)
In the present study, we measured the contact angles for film samples of PC, PC_ATO, and PC_FLU (Table 2). Representative images of the droplets are shown in Supplementary Fig. S5. A common way to measure the contact angle for a solid is to place a liquid droplet on its surface. However, we evaluated the contact angle using films prepared by dissolving the formulations, because we were concerned that powder surface properties such as roughness and porosity might seriously affect the results for the contact angle. Since contact-angle measurements for powders are not reproducible due to the roughness and porosity of the surface, some researchers have used a method involving compressing and molding the powder.28,44) For example, Dahlberg et al. measured the dynamic contact angle for solid dispersions by compression-molding the powders.44) However, it is possible that following compression-molding, the surface roughness and pore structure may depend on the type of powder, making it difficult to compare different powder samples. In addition, it is possible that compression-molding may change the chemical composition of the powder, which would affect droplet adsorption and diffusion. There have been several reports on contact-angle measurements on film samples. Uyama et al. performed contact-angle measurements using polymer film samples formed from solutions cast on glass plates.45) Kwaśniewska et al. investigated the surface properties of starch/powdered activated carbon composite films prepared by the solvent-casting method, including the contact angle.46) Although using a film formed by dissolving powder in a solvent may not allow highly accurate contact-angle measurements, we considered it important to evaluate the effect of the powder composition on the contact angle without the influence of surface irregularities and small pores that are present in the powder. For the above reasons, we therefore evaluated the contact angle using films prepared from the powder formulation instead of the powder itself. However, the effects of particle curvature on the contact angle were not taken into account.
| Film name | Contact angle (dissolution medium) | Contact angle (water) |
|---|---|---|
| PC film | 46.42 ± 3.92° | 77.04 ± 1.80° |
| PC_ATO film | 29.32 ± 1.19° | 52.14 ± 4.14° |
| PC_FLU film | 22.47 ± 1.93° | 21.14 ± 0.98° |
Data are represented as average ± standard deviation (S.D.) (n = 3).
The crystal form of PC obtained by dissolving raw PC crystals in ethanol followed by air-drying on glass was confirmed to be form I in preliminary tests. As shown in Table 2, the contact angle between the PC film and water was 77.04 ± 1.80° due to the high hydrophobicity of PC. In addition, because of the presence of Tween 80 in the dissolution medium, its contact angle is lower than that for water. The contact angle between the dissolution medium and the SD_PC_FLU film was 22.47 ± 1.93°, which was the smallest among those for the measured samples. Therefore, the wettability of SD_PC_ATO and SD_PC_FLU was improved compared with that of the raw PC crystals; SD_PC_FLU, in particular, should exhibit the most hydrophilic properties among the samples. The improved wettability was also a factor in the high PC dissolution rate and high PC release rate from SD_PC_FLU18) (Fig. 5).
Stability of Amorphous StateStability tests were conducted at 40 °C and 75% RH, and the crystallinity of samples was evaluated by PXRD and DSC after 14 d. The PXRD pattern for SD_PC_FLU after 14 d is shown in Fig. 6(I). The PXRD pattern for SD_PC_FLU immediately after preparation showed no diffraction peaks [Fig. 2(e)], whereas peaks were observed in the pattern obtained 14 d after sample preparation [Fig. 6(I)]. In addition, an endothermic peak was detected in the DSC thermogram for SD_PC_FLU stored for 14 d [Fig. 6(II)]; the melting onset point was 109.69 °C. Therefore, SD_PC_FLU could not maintain its amorphous state during storage under such conditions for a period of 14 d. As shown in Fig. 3(II) (e), SD_PC_FLU on the day of sample preparation was almost amorphous, but small amounts of PC forms I and II were detected. The melting onset point for the stored sample was lower than that for SD_PC_FLU (113.45 and 122.98 °C) on the day the sample was prepared [Fig. 3(II) (e)]. By contrast, the peak melting temperature for stored SD_PC_FLU (116.66 °C) was not significantly different from that for SD_PC_FLU (115.47 °C) before the sample was stored. In addition, the melting enthalpy for stored SD_PC_FLU was 20.77 J/g, which is higher than that for SD_PC_FLU on the day the sample was prepared (1.66 and 0.37 J/g). From the above, PC partially existed as form II crystals in SD_PC_FLU, and a small amount of form II PC acted as seed crystals that promoted the crystallization of PC during storage. We have previously conducted stability tests for SD_PC_ATO under the same conditions used in the present study.18) We found that the PXRD pattern for SD_PC_ATO stored for 35 d showed a halo pattern and that no endothermic peak derived from melting of PC was detected in the DSC thermogram.18) Therefore, the amorphous state of SD_PC_ATO was more stable than that of SD_PC_FLU.
Dengale et al. provided a summary of relaxation phenomena that occur in the amorphous state, and reported that molecular motion takes place even at temperatures well below Tg, leading to eventual crystallization into a thermodynamically stable form.37) They pointed out that one of the disadvantages of polymer-based glass solutions, which are classified as being similar to a co-amorphous phase, is the hygroscopicity of polymers. Absorbed moisture can act as a plasticizer that reduces Tg and increases molecular mobility, leading to phase separation and recrystallization. Similarly, it can be assumed that absorbed moisture causes an increase in molecular mobility in an amorphous sample, which leads to more pronounced relaxation and makes the sample more susceptible to crystallization. Although no polymer was used in the present study, the co-amorphous sample containing FLU, which has high wettability, was found to exhibit higher hygroscopicity than SD_PC_ATO, as the appearance of the sample changed when stored in a humid environment [Supplementary Figs. S6(a), S6(b)]. This is expected to increase the molecular mobility in the SD_PC_FLU sample, causing strong relaxation and subsequent crystallization. From the results for drug dissolution from co-amorphous materials, Jensen et al. considered that if the solubility differences between the drug and the co-former are too large, a co-former with a higher solubility may dissolve faster, such that molecular interactions between the drug and co-former may not be strong enough to persist during the dissolution process. In this case, the amorphous drug loses its co-former and is more prone to recrystallization during the dissolution process.40) In the present study, the results of the solubility tests showed that there is a large difference between the solubility of ATO and FLU, which implies that the difference in solubility between PC and FLU is much larger than that between PC and ATO. In addition, the DSC and FTIR spectroscopy results obtained in the present and previous studies showed that new molecular interactions, such as hydrogen bonding, are induced between PC and ATO in co-amorphous PC–ATO, whereas no significant interactions could be identified between PC and FLU in SD_PC_FLU. In the stability tests, the co-amorphous sample of SD_PC_FLU recrystallized faster than that of SD_PC_ATO, which indicates that SD_PC_ATO, in which ATO interacts with PC, is more stable than SD_PC_FLU, where no interaction occurs. Therefore, the solubility of the co-former and the difference in solubility between it and the drug are likely to affect the stability of the co-amorphous samples. On the other hand, the dissolution of PC in freshly prepared SD_PC_FLU was significantly faster and the PC was more soluble than in freshly prepared SD_PC_ATO, which is not necessarily consistent with the results of Jensen et al.40)
There have been other reports that hydrogen bonding contributes to the stability of the amorphous state in solid dispersions. Shibata et al. prepared solid dispersions using 20 kinds of compounds and crospovidone (CrosPVP) and evaluated their properties.47) They reported that all compounds with hydrogen–bond–donor functional groups remained in an amorphous state for at least 6 months after being produced of solid dispersions, and infrared spectra suggested the occurrence of interactions between the functional groups of these compounds and the amide carbonyl groups of CrosPVP. On the other hand, they found that compounds without hydrogen-bond-donor groups could not maintain an amorphous state and recrystallized within one month. Therefore, they considered that the presence of such functional groups is an important factor influencing the formation of stable solid dispersions with CrosPVP containing hydrogen bond acceptors.47) Based on the results of our previous research, new molecular interactions such as hydrogen bonding may occur between the carbonyl and amide groups of PC and ATO, and hence a phenomenon similar to that described above may be occurring, although there is a difference between a solid dispersion and a co-amorphous state.18) Therefore, it is considered that hydrogen bonds improve the stability of the amorphous state.
On the other hand, as described in the Thermal analysis and FTIR spectroscopy sections, co-amorphization can improve amorphous stability even when no significant molecular interactions are present. Löbmann et al. reported that a co-amorphous phase formed by SVS and GPZ showed improved stability that was attributable to molecular-level mixing.38) Dengale et al. summarized the results of recent studies on co-amorphous systems and pointed out that in most cases, the physical stability can be attributed to molecular interactions, and also that improved stability was observed as a result of intimate mixing in some co-amorphous systems.37) As described above, in our present and previous study, molecular interactions such as hydrogen bonding were found to occur in the SD_PC_ATO sample, but no clear interactions were identified for SD_PC_FLU, whose physical stability is considered to have been improved by molecular-level mixing. The stability tests in the present study determined that SD_PC_ATO can maintain an amorphous state longer than SD_PC_FLU. This high stability indicates that molecular interactions can significantly hinder recrystallization. It should be noted that although we primarily focus on molecular interactions and molecular-level mixing in the present study, there may also be other factors that affect the stability of the amorphous state.
Release profiles for PC from samples of SD_PC_FLU and SD_PC_ATO stored for 14 d at 40 °C and 75% RH are shown in Fig. 7. The percentage of PC dissolved from stored SD_PC_FLU was 1.74 ± 0.60%, which is drastically smaller than that before storage (Figs. 5, 7). As previously described, in the case of SD_PC_FLU stored for 14 d, diffraction peaks were detected in the PXRD pattern and an endothermic peak originating from the melting of PC form II was observed in the DSC thermogram (Fig. 6). Therefore, the drastic decrease in the percentage of PC dissolved from the stored SD_PC_FLU is attributed to recrystallization of PC in the sample during storage. However, the percentage of PC dissolved from stored SD_PC_FLU was still higher than that dissolved from PM_PC_FLU and raw PC crystals. In the case of PM_PC_FLU, diffraction peaks originating from PC and FLU were detected in the PXRD pattern and an endothermic peak originating from the melting of PC form I was observed in the DSC thermogram [Figs. 2(b), 3(b)]. As previously noted, the crystal form of PC in the raw PC crystals and PM_PC_FLU was identified as form I (stable form) and that in the stored SD_PC_FLU was identified as form II (metastable form).18) Therefore, we speculate that one of the reasons for the difference in the percentage of dissolved PC between SD_PC_FLU and PM_PC_FLU after storage is a difference in crystal form. It is also possible that some fraction of the amorphous phase remained in SD_PC_FLU after storage, resulting in a higher percentage of dissolved PC. In contrast to the percentage of PC dissolved from SD_PC_FLU, that dissolved from the stored SD_PC_ATO was 14.16 ± 0.89% at 2 h (Fig. 7). As noted in the section on Dissolution tests, we previously reported that freshly prepared SD_PC_ATO showed approx. 9% drug release after 2 h.18) Surprisingly, the percentage of PC dissolved from SD_PC_ATO increased after storage at 40 °C and 75% RH. We also previously reported that the co-amorphous state of SD_PC_ATO could be maintained for 35 d without recrystallization.18) Therefore, we propose that the solubility of the PC in SD_PC_ATO did not decrease after storage because the PC was maintained in an amorphous state. In addition, the morphology of stored SD_PC_ATO particles did not change after preparation, whereas aggregated and fused particles were observed for SD_PC_FLU after storage [Supplementary Figs. S6(a), S6(b)]. These results indicate that SD_PC_ATO exhibits greater storage stability than SD_PC_FLU. Unfortunately, the reason for the increased solubility of PC in SD_PC_ATO after storage remains unclear and is left as a topic for future work.
Data are represented as average ± S.D. (3 ≤ n ≤ 5). ■, SD_PC_FLU; ▲, SD_PC_ ATO.
In the present study, a co-amorphous state formed by PC and FLU was prepared by a SD method (SD_PC_FLU). We previously reported that the solubility of PC is improved by forming a co-amorphous state with ATO via SD (SD_PC_ATO). We have also reported that the co-amorphous state of SD_PC_ATO can be maintained for 35 d without recrystallization.18) In the present study, SD_PC_FLU and SD_PC_ATO were compared in terms of their physicochemical properties and storage stability under accelerated test conditions of 40 °C and 75% RH. The PC in SD_PC_FLU exhibited a higher dissolved percentage and a higher dissolution rate than that in SD_PC_ATO. We attributed these effects to FLU exhibiting greater solubility and wettability than ATO. Regarding the difference in physicochemical properties between SD_PC_FLU and SD_PC_ATO, FTIR measurements indicated that new molecular interactions occurred in SD_PC_ATO but not in SD_PC_FLU. In the stability tests, diffraction peaks and an endothermic peak derived from PC crystals (form II) appeared in the PXRD pattern and DSC thermogram, respectively, for SD_PC_FLU stored for 14 d under the accelerated test conditions. Therefore, the stability of the amorphous state of PC in SD_PC_ATO is higher than that in SD_PC_FLU. This enhanced stability is attributed to several factors, including the physicochemical properties of the co-former, such as its solubility, and the different molecular interaction modes between PC–FLU and PC–ATO. In addition, the percentage of PC dissolved from SD_PC_FLU after storage drastically decreased, whereas no decrease was observed for SD_PC_ATO after storage.
In conclusion, the present study shows that, when a more polar molecule is selected as a co-former for the co-amorphous state, the co-amorphous wettability is improved and the solubility-improving effect can be enhanced. The results of the present study also indicate that molecular interactions enhance the stability of the co-amorphous state. We found that the use of FLU and ATO, which have similar chemical structures and pharmacological effects, as co-formers for PC resulted in a high solubility of PC and high stability of the amorphous state, respectively. In the future, further improvements are expected to lead to a formulation design that combines the better dissolution of PC and greater stability of the co-amorphous state.
This study was supported by a Kakenhi Grant-in-Aid (No. 18K06610) from the Japan Society for the Promotion of Science (JSPS).
Shinji Oyama: Conceptualization, Methodology, Investigation, Validation, Writing—Original Draft. Noriko Ogawa: Conceptualization, Methodology, Investigation, Validation, Writing—Original Draft, Funding acquisition. Toshiya Yasunaga: Writing—Review & Editing. Hiromitsu Yamamoto: Supervision, Methodology, Writing—Review & Editing, Funding acquisition.
The authors declare no conflict of interest.
This article contains supplementary materials.
