Unravelling the Relationship between Degree of Disorder and the

Nov 13, 2013 - used to characterize degree of disorder that correlates best with the recrystallization behavior during dissolution of milled glibencla...
4 downloads 5 Views 5MB Size
Article pubs.acs.org/molecularpharmaceutics

Unravelling the Relationship between Degree of Disorder and the Dissolution Behavior of Milled Glibenclamide Pei T. Mah,†,‡ Timo Laaksonen,‡ Thomas Rades,§ Jaakko Aaltonen,‡ Leena Peltonen,*,‡ and Clare J. Strachan‡ †

School of Pharmacy, University of Otago, Dunedin, New Zealand Division of Pharmaceutical Technology, Faculty of Pharmacy, University of Helsinki, Finland § Department of Pharmacy, University of Copenhagen, Denmark ‡

ABSTRACT: Milling is an attractive method to prepare amorphous formulations as it does not require the use of solvents and is suitable for thermolabile drugs. One of the key critical quality attributes of milled amorphous formulations is their dissolution behavior. However, there are limited studies that have investigated the relationship between degree of disorder induced by milling and dissolution behavior. The main aim of this study was to identify the analytical technique used to characterize degree of disorder that correlates best with the recrystallization behavior during dissolution of milled glibenclamide samples. Solid state and surface changes during dissolution of milled glibenclamide samples were monitored in order to elucidate the processes that influence the dissolution behavior of milled glibenclamide samples. Glibenclamide was ball milled for different durations and analyzed using X-ray powder diffractometry (XRPD), Raman spectroscopy and differential scanning calorimetry (DSC). Recrystallization during dissolution of the milled amorphous materials was investigated using an in situ Raman setup. SEM was used to monitor the surfaces of the compacts during dissolution. XRPD, Raman spectroscopy and DSC indicated that glibenclamide was fully amorphous after milling for 30, 60, and 120 min, respectively. ‘DSC amorphous’ (i.e. fully amorphous according to the onset of crystallization obtained from DSC) glibenclamide samples experienced negligible recrystallization which had no effect on the dissolution profiles. Samples that were not ‘DSC amorphous’ experienced recrystallization which resulted in a decrease in dissolution rate. Unexpected elevated dissolution rate was observed initially during dissolution for samples milled for 15 to 45 min, and this was related to particle loss from surfaces of the disks during dissolution. In conclusion, the onset of crystallization obtained from DSC best predicts the recrystallization of glibenclamide during dissolution. Recrystallization and particle loss from the surface of the dissolution should be considered when interpreting the dissolution data of milled glibenclamide samples. KEYWORDS: amorphous, crystallinity, dissolution, glibenclamide, milling, recrystallization, XRPD, Raman spectroscopy, DSC, solid state analysis



INTRODUCTION

during processing, storage and dissolution. This will negate its solubility advantage and may subsequently lead to therapeutic failure.5−7 Different preparative methods have been used to formulate drugs into the amorphous form, and these include milling, melt extrusion, spray drying and freeze-drying.7,8 Milling is an attractive method to prepare amorphous formulations as it does not require the use of solvents and is suitable for thermolabile drugs.9 The use of this method is not limited to preparing single-component amorphous systems. Solid dispersion comprising drug/polymer systems and binary co-amorphous systems which consist of drug/drug or drug/amino acid

The majority of the drugs under development are poorly water soluble and often have a low dissolution rate and ultimately poor oral bioavailability. It is a challenge to formulate poorly water-soluble drugs into oral formulations, and such drugs account for a large proportion of the pharmaceutical products on the market. This has driven efforts to develop strategies to formulate these drugs, which include the use of cosolvents, cyclodextrin complexes, lipid-based drug delivery systems (e.g., solid lipid nanoparticles or self-emulsifying drug delivery systems) and reduction of drug particle size.1−4 Formulating drugs into an amorphous form is a promising approach to improve the aqueous solubility of drugs.5,6 The amorphous form does not have long-range order and has a higher apparent solubility than its ordered, crystalline counterpart. However, the amorphous form is not thermodynamically stable and has a tendency to revert back to the crystalline form © 2013 American Chemical Society

Received: Revised: Accepted: Published: 234

July 17, 2013 October 25, 2013 November 13, 2013 November 13, 2013 dx.doi.org/10.1021/mp4004145 | Mol. Pharmaceutics 2014, 11, 234−242

Molecular Pharmaceutics

Article

systems have been successfully fabricated by milling.9−15 Currently, amorphization via milling has been proposed to occur either via progressive structural disordering or directly via quench melting due to elevated local temperature generated during the milling process.16,17 There is a substantial amount of research focusing on investigating factors that affect the physical stability of amorphous formulations during storage18,19 and predicting physical stability.20−24 Different preparative methods and processing parameters have been shown to influence the physical stability of amorphous materials upon storage.19,25,26 Bøtker et al. analyzed indomethacin cryomilled for different durations using various analytical techniques. They found that indomethacin was X-ray amorphous after milling for 30 min, while ssNMR and DSC suggested that indomethacin was fully amorphous after 195 min of milling.26 The differences in the indomethacin milled for various durations were attributed to either different degrees of disorder or amounts of crystalline nuclei, and these differences affected the physical stability of cryomilled indomethacin upon storage. This study demonstrated that different analytical techniques have different sensitivity to crystallinity and found that ssNMR and DSC predicted the physical stability of indomethacin best.26 At present, there are few studies that have investigated the effect of differences in degree of disorder or amount of nuclei as a result of different milling duration on the recrystallization behavior during dissolution. Karmwar et al. found that the dissolution rate of cryomilled indomethacin increased with increasing milling duration.27 However, this study did not correlate the signal of analytical techniques used to characterize the cryomilled samples to the dissolution behavior. The recrystallization process that occurred during dissolution of the samples cryomilled for various durations was also not investigated. The understanding of recrystallization that occurs during dissolution of amorphous formulations is pivotal for the development of marketable amorphous formulations. The model drug used in this study was glibenclamide (Figure 1) which is a sulphonylurea-based antidiabetic drug and is

critical quality attribute of milled amorphous formulations. Surface changes during dissolution of milled glibenclamide samples were monitored in order to elucidate the processes that influence the dissolution behavior of milled glibenclamide samples.



MATERIALS AND METHODS Materials. Glibenclamide was purchased from Hangzhou Dayangchem (Hangzhou, China). Boric acid (Sigma Aldrich, Saint Louis, MO, U.S.A.), potassium chloride (Sigma Aldrich, Steinheim, Germany) and sodium hydroxide (Sweden) were used to prepare the buffer for the dissolution experiments. Milling. Milling was carried out using the planetary ball mill Pulverisette 6 (Fritsch GmbH, Idar-Oberstein, Germany). Glibenclamide form I powder (1.5 g) was placed in an 80mL volume stainless steel bowl, containing 15 stainless steel balls with a diameter of 10 mm. The bowl containing the sample was precooled in an ice bath prior to milling. Milling was performed at 400 rpm for 5, 10, 15, 30, 45, 60, 90, 120, 150, and 180 min. The sample in the milling bowl was scraped down from the wall and lid of the bowl after every 15 min of milling to ensure milling homogeneity. The bowl containing the sample was cooled in an ice bath for 15 min after every 15 min of milling to minimize heating. XRPD. The milled glibenclamide samples were analyzed by XRPD with a Bruker D8 Advance system (Bruker AXC GmbH, Karlsruhe, Germany) using Cu Kα radiation with λ = 1.542 Å (40 kV and 40 mA) and a divergence slit of 1°. The samples were gently pressed into flat aluminum sample holders using a spatula and scanned from 5° to 35° 2θ with a scanning rate of 0.1°/s. Raman Spectroscopy. Raman spectra were collected using a Raman Rxn1 system (Kaiser Optical Systems, Ann Arbor, MI, U.S.A.) which was equipped with a 785 nm excitation laser source, a PhAT probe which consisted of an array of 50 optical fibers, and an air-cooled charge-coupled device (CCD) detector. The sampling spot size of this system was 6 mm in diameter, and the size of the area illuminated was 28.3 mm2. The integration time used was 5 s, and the final spectrum was the mean of 5 scans. The spectral resolution was 4 cm−1. Each sample was measured in triplicate. Data collection and conversion was performed using the HoloGRAMS 4.1 (Kaiser Optical Systems) software. DSC. DSC thermograms were recorded using the Mettler DSC 823e (Mettler-Toledo AG, Greifensee, Switzerland). Samples (1−2 mg) were crimped in aluminum pans with pierced lids, equilibrated at 0 °C for 5 min and finally heated up to 200 °C at a heating rate of 10 °C/min. The measurement cell was purged with dry nitrogen gas at a flow rate of 50 mL/ min during the measurements. The glass transition temperature (Tg), onset of crystallization, enthalpy of crystallization and melting temperature were determined using the STARe software (Mettler-Toledo AG, Greifensee, Switzerland). The Tg was defined as the midpoint change in heat capacity of the sample and the melting temperature was defined as the onset of the peak. Each sample was measured in triplicate. Intrinsic Dissolution Testing. Tablet compacts weighing approximately 150 mg were prepared using the Specac Hydraulic Press model 15.011 (Specac, Kent, UK) equipped with a 13 mm diameter flat-faced punch. The compacts were formed with a compaction pressure of 36.9 MPa and a dwell time of 30 s. The intrinsic dissolution studies were performed using a dissolution flow-through cell setup similar to that used

Figure 1. Molecular structure of glibenclamide.

classified as a Biopharmaceutics Classification System (BCS) class II drug (poor aqueous solubility and high permeability).28 There are three reported polymorphic forms.28−30 It has been found that quench cooling of glibenclamide resulted in a significant chemical degradation.31 Milling has been shown to successfully convert glibenclamide into the amorphous form without significant degradation.31 The first part of this study aims to compare XRPD, Raman spectroscopy, and DSC in characterizing pharmaceutical systems with different degrees of disorder as a result of different milling durations. Then, the subsequent aim of this study is to determine the analytical technique that best predicts the recrystallization behavior of the model drug during dissolution and ultimately the dissolution rate. To the best of our knowledge, no studies have correlated the signal of different analytical techniques to dissolution behavior which is a key 235

dx.doi.org/10.1021/mp4004145 | Mol. Pharmaceutics 2014, 11, 234−242

Molecular Pharmaceutics

Article

by Aaltonen et al.32 The flow-through cell included a quartz glass plate to enable in situ Raman probing of the tablet surface during dissolution. The dissolution medium (borate buffer, pH 9, 900 mL) was maintained at a temperature of 37 °C and was stirred at 50 rpm. The flow rate of the dissolution medium was 4 mL/min. The concentration of glibenclamide in the dissolution medium was monitored every 60 s for a duration of 60 min using a UV−vis spectrophotometer (UV-1600PC Spectrophotometer, VWR, China) at the wavelength of 230 nm. The dissolution studies were performed in triplicate. The solid state changes of the tablet surfaces during dissolution were monitored by placing the Raman probe above the surface of the tablet at a distance of approximately 23 cm. The Raman system and parameters used were the same as described in the Raman Spectroscopy section. Raman Data Analysis. The Raman spectral region between 764 and 1288 cm−1 was included in the analysis. The spectra were subsequently subjected to baseline offset and linear baseline correction to remove baseline differences. Finally, the spectra underwent standard normal variate (SNV) transformation in order to remove the remaining spectral differences that were not related to the sample composition. The preprocessed spectra were then subjected to principal component analysis (PCA) to investigate the differences in the samples. The spectral preprocessing and PCA were performed using the Unscrambler X software (v. 10.1, Camo Software AS, Norway). SEM. Scanning electron microscopy images of the tablet surfaces were recorded using the FEI Quanta 250 FEG (FEI Inc., Eindhoven, The Netherlands) scanning electron microscope equipped with the Everhart-Thornley detector (ETD). The working voltage of 10 kV was used. The tablet surfaces were gently dried using tissue paper. The tablets were then mounted on aluminum stubs with double-sided carbon tape and subsequently were coated with platinum before imaging.

Figure 2. XRPD patterns of glibenclamide milled for different durations. The times in the figure indicate the milling durations.



RESULTS AND DISCUSSION Characterization of Milled Glibenclamide Samples. XRPD. The XRPD patterns of glibenclamide milled for different durations are shown in Figure 2. The diffraction pattern of the starting material corresponded to that of the Cambridge Structural Database (CSD) reference for glibenclamide form I (reference code: DUNXAL).33 The X-ray diffractograms of glibenclamide samples milled for 15 min or less exhibited peaks characteristic of crystalline glibenclamide with a halo background, indicating that these samples were partially amorphous, containing some residual crystallinity. Milling glibenclamide for 30 min or longer resulted in samples with diffractograms that had a broad halo and an absence of diffraction peaks. This suggested that glibenclamide samples milled for 30 min or longer were ‘X-ray amorphous’. The evolution of the diffractograms during milling did not suggest that amorphization occurred through an intermediate form. Raman spectroscopy. The changes that occurred in the Raman fingerprint region of glibenclamide during milling are depicted in Figure 3. Some vibrational modes of functional groups have been tentatively assigned to certain peaks by comparing the Raman spectra of glibenclamide to that of another sulphonylurea drug, chlorpromamide in which Raman bands have previously been assigned.34 The peak at 1714 cm−1 is likely to correspond to CO stretching. This peak disappeared upon amorphization which most probably is associated with changes in hydrogen bonding. The intensity

Figure 3. Raman spectra of glibenclamide milled for different durations. The times in the figure indicate the milling durations. The gray regions are regions with major spectral changes occurring during milling.

236

dx.doi.org/10.1021/mp4004145 | Mol. Pharmaceutics 2014, 11, 234−242

Molecular Pharmaceutics

Article

of the shoulder of the peak at 1593 cm−1 increased upon amorphization. This peak is attributed to CC stretching of the benzene rings. The doublet at 1345 cm−1 and the peak at 1156 cm−1 which have been assigned to asymmetric and symmetric SO2 stretching modes, respectively, broadened upon amorphization. Futhermore, amorphization resulted in merging of the multiplet at 1442 cm−1, which could be assigned to deformation vibrations of the alkyl groups. There were also minor changes in the bands within the region of 400−700 cm−1 which could be associated with in-plane deformation of the aromatic ring and the SO 2 group. Amorphization of glibenclamide also resulted in a peak shift from 811 to 803 cm−1, reductions in peak intensity at 1020, 1182, and 1202 cm−1, and peak broadening at 1249 cm−1. Peak broadening and reduction in peak intensities are commonly observed in the transformation from the crystalline to the amorphous form.35−37 The Raman data of the milled glibenclamide samples were subjected to principal component analysis (PCA) to separate the samples in order to determine the milling time required to render glibenclamide fully amorphous. Principal component 1 (PC1) explained 95% of the variation, and the loadings plot of PC1 (Figure 4a) suggests that PC1 described the amorphization path. While the peaks at 783, 800, 1090, and 1274 cm−1 are characteristic of amorphous glibenclamide, the peaks at 812, 824, 842, 1020, 1059, 1097, 1202, and 1249 cm−1 are characteristic of crystalline glibenclamide. This suggests that crystallinity was negatively correlated with PC1 and amorphousness was positively correlated with PC1. Examination of

the loadings plot of other PCs revealed that these did not contain systematic variation. In order to enable easy interpretation of the PCA data, the PC1 values were plotted as a function of milling duration (Figure 4b). The PC1 value increased with increasing milling duration and reached a plateau after 60 min of milling. This indicates that PCA was not able to differentiate samples that were milled 60 min and longer, implying that according to Raman spectroscopy, glibenclamide was fully amorphous after 60 min of milling. DSC. The thermal properties of glibenclamide milled for different durations are presented in Table 1. The thermograms of all the milled samples displayed a glass transition at about 74 °C, a recrystallization exotherm with onset temperatures between 89 and 119 °C and a melting endotherm with onsets between 165 and 169 °C. The presence of a glass transition and a recrystallization exotherm in the thermogram of the glibenclamide after 5 min of milling indicates that the sample was partially amorphous. The values of Tg obtained in this study were similar to those reported in the literature.31 Bimodal recrystallization exotherms were consistently observed in the thermograms of the glibenclamide samples milled for 30, 45, and 60 min (Figure 5). This is possibly due to different crystallization behavior between the surface and the bulk of the milled glibenclamide particles38 or due to transformation via an intermediate form.39 The onset and enthalpy of crystallization of glibenclamide milled for 30, 45, and 60 min recorded in Table 1 were determined from the whole bimodal exotherm. Enthalpy and onset of crystallization increased with increasing milling duration and reached a constant value after milling for 60 and 120 min respectively. Similar observations were also reported previously by Bøtker et al. who cryomilled indomethacin for different durations.26 These observations are likely due to increasing disorder and a reduction in the number of nuclei in the samples with increasing milling duration.26 The enthalpy of crystallization was not a good indicator for degree of disorder in glibenclamide due to the presence of the bimodal exotherm in some of the milled samples which may indicate transformation via an intermediate form. This complicated the process of computing the true enthalpy of crystallization for those samples. Thus, the onset of crystallization was used as the indicator of the degree of disorder for glibenclamide. Milling glibenclamide for 120 min or longer did not cause any apparent changes in the onset of crystallization. This suggests that milling glibenclamide for longer than 120 min did not further increase the degree of disorder and glibenclamide was fully amorphous after 120 min. Comparison of the Techniques to Characterize Degree of Disorder. The above results show that XRPD, Raman spectroscopy and DSC suggest that glibenclamide was fully amorphous after different milling durations (i.e., 30, 60, and 120 min, respectively). These results support that amorphization via milling is a progressive structural disorder process and the degree of crystallinity of a sample changes on a continuum scale, ranging from 0 to 100%. The degree of crystallinity depends on the degree of lattice disorder and cannot be described by binary mixtures of 100% amorphous and 100% crystalline material. Milling induces localized defects to the crystal structure and these crystal defects accumulate as milling progresses until the sample achieves cooperative disorder (i.e. amorphous).40

Figure 4. (a) PC1 loadings plot of the Raman data of milled glibenclamide samples. (b) PC1 values of milled glibenclamide samples as a function of milling duration. 237

dx.doi.org/10.1021/mp4004145 | Mol. Pharmaceutics 2014, 11, 234−242

Molecular Pharmaceutics

Article

Table 1. Thermal Properties of Glibenclamide Milled for Different Durations (mean ± SD, n = 3) milling duration (min) 0 5 10 15 30 45 60 90 120 150 180

midpoint of glass transition, Tg (°C) − 75.83 75.60 75.42 74.62 73.53 74.23 73.50 74.88 74.77 74.50

± ± ± ± ± ± ± ± ± ±

onset of crystallization (°C)

1.49 0.91 0.26 0.99 1.45 0.85 1.34 1.01 2.50 1.15

− 89.86 92.34 94.46 96.30 99.26 102.69 106.47 115.51 118.73 119.43

± ± ± ± ± ± ± ± ± ±

0.78 0.23 0.27 0.12 0.22 0.85 1.11 2.51 1.49 0.83

enthalpy of crystallization (J/g) − 23.88 36.81 40.72 41.31 41.60 57.17 58.78 59.29 59.75 61.25

± ± ± ± ± ± ± ± ± ±

4.93 3.53 3.63 0.60 3.75 2.00 2.41 2.57 2.70 3.78

onset of melting (°C) 172.77 168.54 167.71 167.50 165.54 166.24 166.86 169.39 169.54 169.47 169.49

± ± ± ± ± ± ± ± ± ± ±

0.16 0.38 0.24 0.63 1.51 1.11 0.19 0.50 0.47 0.48 0.63

duration of milling that is required in order to obtain the greatest dissolution benefits. The dissolution profiles of the glibenclamide samples milled for different durations are illustrated in Figure 6.

Figure 6. Dissolution profiles of glibenclamide milled for various durations. The times in the legend indicate the milling durations. The terms of ‘X-ray amorphous’, ‘Raman amorphous’ and ‘DSC amorphous’ are used to describe samples that were fully amorphous according to XRPD, Raman spectroscopy and DSC, respectively, in this study. Error bars (all were