Manufacturing Amorphous Solid Dispersions with a Tailored Amount

Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Manufacturing Amorphous Solid Dispersions with a Tailored Amount of Crystallized API for Biopharmaceutical Testing Frank Theil, Johanna Milsmann, Sankaran Anantharaman, and Holger van Lishaut* AbbVie Deutschland GmbH & Co. KG, 67061 Ludwigshafen, Germany S Supporting Information *

ABSTRACT: The preparation of an amorphous solid dispersion (ASD) by dissolving a poorly water-soluble active pharmaceutical ingredient (API) in a polymer matrix can improve the bioavailability by orders of magnitude. Crystallization of the API in the ASD, though, is an inherent threat for bioavailability. Commonly, the impact of crystalline API on the drug release of the dosage form is studied with samples containing spiked crystallinity. These spiked samples possess implicit differences compared to native crystalline samples, regarding size and spatial distribution of the crystals as well as their molecular environment. In this study, we demonstrate that it is possible to grow defined amounts of crystalline API in solid dosage forms, which enables us to study the biopharmaceutical impact of actual crystallization. For this purpose, we studied the crystal growth in fenofibrate tablets over time under an elevated moisture using transmission Raman spectroscopy (TRS). As a nondestructive method to assess API crystallinity in ASD formulations, TRS enables the monitoring of crystal growth in individual dosage forms. Once the kinetic trace of the crystal growth for a certain environmental condition is determined, this method can be used to produce samples with defined amounts of crystallized API. To investigate the biopharmaceutical impact of crystallized API, non-QC dissolution methods were used, designed to identify differences between the various amounts of crystalline materials present. The drug release in the samples manufactured in this fashion was compared to that of samples with spiked crystallinity. In this study, we present for the first time a method for targeted crystallization of amorphous tablets to simulate crystallized ASDs. This methodology is a valuable tool to generate model systems for biopharmaceutical studies on the impact of crystallinity on the bioavailability. KEYWORDS: amorphous solid dispersion, crystal growth, transmission Raman spectroscopy, biopharmaceutical testing

1. INTRODUCTION More and more poorly water-soluble active pharmaceutical ingredients (APIs) are emerging from discovery organizations of pharmaceutical companies. The main challenge in the development of solid oral dosage forms with these APIs is the lack of solubility in aqueous systems.1,2 An elegant way to overcome this challenge, enhance the water solubility, and therefore potentially enhance bioavailability is the formulation as an amorphous solid dispersion (ASD).3,4 In the case that solubility is the rate-limiting step (biopharmaceutical classification system (BCS) II and to a certain extent BCS IV drugs), this formulation strategy has been proven to successfully enhance the bioavailability of numerous APIs.5 In ASD formulations, the API is stabilized in its amorphous form by molecular interactions with the polymer-like hydrogen bonds and van der Waals forces.6 Even though the amorphous form of the API is stabilized in an ASD, the crystalline form of a drug substance remains the thermodynamically stable solid form. The amorphous phase is therefore prone to recrystallization over time. An important factor determining the rate of this phase transformation is the drug load of the ASD, and therefore it is one of the key factors impacting the physical stability of the © XXXX American Chemical Society

formulation. Nevertheless, a drug load as high as possible in an oral dosage form is highly desirable to lower the pill burden for the patients.6 Hence, exploring the limits of stabilization for high drug load ASD formulations to avoid recrystallization is a challenging task for both pharmaceutical research and industry.7,8 Since crystalline API may impact the bioavailability of the dosage form, crystallinity is considered a critical quality attribute in ASD production.9 In principal, crystallinity can emerge in two ways in an ASD product: as residual crystallinity originating from the production process and as crystallized API emerging during the storage of the product.10 Residual crystallinity stems from crystalline API that is not completely dissolved in the polymer matrix during production of the ASD.11 This form of API crystallinity is well excluded by commonly applied quality control procedures and is generally not perceived as a risk for patients.12 Crystallized API stems Received: January 15, 2018 Revised: April 3, 2018 Accepted: April 4, 2018

A

DOI: 10.1021/acs.molpharmaceut.8b00043 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

nondestructive method to determine crystalline content in whole solid dosage forms. As a fast crystallizer and BCS II compound, fenofibrate is used in this study. The crystal growth in a copovidone-based fenofibrate formulation is investigated, and the production of tablets with a defined crystal content of fenofibrate is demonstrated. The tablets produced in this fashion were then used to determine the impact of crystal growth on API bioavailability. As the kinetic trace of crystal growth in ASD formulations follows the same physics, regardless of API or excipients, the methodology is not limited to copovidone-based fenofibrate formulations, but it is generally applicable to any fast crystallizing ASD.

from crystal growth in the ASD formulations during the storage of the product and can occur only in such formulations that are oversaturated solid solutions in a state of kinetical stabilization. 13 Fortunately, even a kinetically stabilized ASD formulation can be stable over decades.14−17 The crystallization of API is therefore also not a priori a risk in the currently marketed pharmaceutical products. However, the resilience of a kinetically stabilized ASD to crystallize is strongly dependent on the molecular mobility in the formulation and, hence, the difference between the storage temperature and the glass transition temperature Tg. The Tg itself is, among other factors, dependent on the drug load. In most cases, a higher drug load causes a lower Tg, as the Tg of the neat API is often below the Tg of the used polymer. Since drug loads as high as physically possible for ASD products are nevertheless strongly desired by the pharmaceutical manufacturers to lower the pill burden for patients,6 exploring the boundaries of physical stabilization is a necessity for the pharmaceutical industry. For products that exhibit only borderline physical stability, it is imperative to be able to simulate recrystallized material and investigate biopharmaceutical impact. A crystallized ASD potentially has physicochemical properties different than those of a spiked sample, which might imply a change in the dissolution behavior. The size distribution of the grown crystals, the polymorph, and the spatial distribution can indeed be different than those in a sample spiked with crystalline API.18 Additionally, the microenvironment of the crystals is changed, as the growth of crystals in the formulation will deplete the surrounding material of API (Figure 1).

2. METHODS 2.1. Chemicals. Copovidone was purchased from BASF SE (Ludwigshafen am Rhein, Germany), fenofibrate from Midas Pharma GmbH (Ingelheim am Rhein, Germany), Labrafil from Gattefossé GmbH (Bad Krozingen, Germany), and Aerosil 200 from Evonik Degussa GmbH (Essen, Germany). KH2PO4, K3PO4·H2O, H3PO4, Na2HPO4, NaH2PO4·2H2O, cetyltrimethylammmonium bromide (CTAB), and sodium dodecyl sulfate (SDS) were purchased from Sigma-Aldrich (Steinheim, Germany). Acetonitrile (LiChrosolv) for HPLC measurements was purchased from Merck KGaA (Darmstadt, Germany). All other organic solvents were of analytical grade and were purchased from Merck KGaA (Darmstadt, Germany). 2.2. Sample Preparation. The manufacturing of the fenofibrate formulation was already discussed in our earlier work.16 In brief, a blend containing 15% (w/w) fenofibrate, 82% (w/w) copovidone K28, 2% (w/w) Labrafil, and 1% (w/ w) colloidal silica was extruded using a Micro 18 twin-screw extruder (Leistritz AG, Nürnberg, Germany). A placebo extrudate was prepared in the same manner. The tablets were prepared by mixing 97.6% (w/w) milled extrudate beads with 1.1% (w/w) Aerosil 200 and 1.3% (w/w) sodium stearyl fumarate. This mixture was used to compress football-shaped tablets with a single-punch tablet press (Ek0, Korsch AG, Berlin, Germany). For tablets with spiked API content, part of the amorphous API content was replaced with crystalline API, and a placebo extrudate was used to adjust for the missing extrudate matrix material. The extrudate and placebo extrudate particles as well as the API had a D50 values of 25, 271, and 213 μm, respectively, as measured by the Mastersizer 3000 (Malvern Instruments, Malvern, U.K.). 2.3. Assay, Degradation, and Water Content. As discussed in our previous work,16 HPLC measurements were performed on a Waters 2695 separation instrument. The detection wavelength was set to 286 nm. The eluent used was a binary mixture of acetonitrile and aqueous KH2PO4-buffer solution (pH = 3.0) in a linear gradient from 38 to 60% acetonitrile. The HPLC data were evaluated by area and quantified against an external standard. The water content of the samples was measured by Karl Fischer titration using a Titrino 784 (Deutsche Metrohm, Filderstadt, Germany). Methanol was used as a solvent and titrated to dryness with hydranal composite 5 before each measurement. 2.4. PLM, DSC, and XRPD. The physical state of the samples was investigated by polarized light microscopy (PLM), differential scanning calorimetry (DSC), and X-ray powder diffraction (XRPD). A detailed description of the experimental methods can be found elsewhere.16 PLM images were acquired using a DMLM optical microscope (Leica Microsystems, Wetzlar, Germany). DSC experiments were performed on a

Figure 1. Scheme of the two types of potential crystallinity in ASDs and depiction of the anticipated physical effects.

It is of course not possible to simulate crystallized material in an ASD formulation that is inherently resilient to crystallization, nor is it necessary, as the threat of crystallization during the storage period is not given. In other cases though, where an unusually high drug load is desired to achieve sufficient bioavailability and a tendency to crystallize is observed in the particular formulation, the investigation of the biopharmaceutical impact of a crystallized ASD is recommended. In this article, we present a method to grow predefined amounts of crystalline API in individual solid dosage forms for investigation in biopharmaceutical in vitro or even in vivo tests. Transmission Raman spectroscopy (TRS) in combination with chemometric data evaluation is applied as a fast and B


Article


the short measurement time of TRS has no influence on the crystal growth in the samples. The time needed to acquire the spectrum is orders of magnitude shorter than the overall storage period. In the following, we first demonstrate the feasibility of TRS to quantify crystals in the investigated fenofibrate formulation. The crystal growth in tablet samples is then successfully followed with the developed spectroscopic method in individual dosage forms stored under stress conditions of 40 °C and 75% RH. The same conditions are then used to repeat the crystal growth experiment. For the production of samples with tailored crystal content, less harsh conditions were chosen to achieve better control over the grown crystalline content. For this purpose, tablets of the investigated fenofibrate formulation were stored at 25 °C and 76% RH as well as 80% RH. The kinetic trace of crystal growth at these particular conditions was first studied until the phase equilibrium was reached to explore the progress of the phase transition. Afterward, the acquired rate of phase transition was applied to produce samples with a defined amount of crystalline content. The biopharmaceutical impact of the crystalline content in these samples was then investigated by dissolution testing. 3.1. Initial Characterization of the Fenofibrate Formulation. To ensure the quality of the investigated fenofibrate formulation content and purity, water content, glass transition temperature, as well as the crystallinity were investigated (Table 1). The content of extrudate and tablets

DSC1 (Mettler-Toledo, Giessen, Germany). XRPD measurements were performed on an X’pert Pro MPD system (PANalytical, Almelo, The Netherlands). 2.5. Crystal Growth Investigation. TRS was used to investigate crystal growth, as the rapid and nondestructive determination of crystallinity in the samples by TRS makes it possible to follow the crystal growth in individual dosage forms. The TRS spectra were measured with a TRS100 spectrometer (Agilent, Oxford, U.K.). The TRS100 uses area illumination optics. The laser illumination spot size was set to an 8 mm diameter, and small lens collection optics were used. To quantify the content of the crystalline fenofibrate multivariate, analysis of the TRS spectra was conducted. The Unscrambler X 10.4 software package (CAMO Software, Oslo, Norway) was used to develop the model and quantify the crystalline content from the acquired spectral data. The experimental details and chemometric method development are documented in a previous publication.16 To manufacture tablets with defined amounts of crystallized API, the crystal growth traces of the investigated formulation at the defined environmental conditions need to be recorded. The manually pressed tablets (500 mg, 13 mm diameter) were placed in sampling trays for the TRS100 instrument. For the crystal growth experiments, sample trays were directly placed in climate chambers. In order to demonstrate the feasibility of this approach, a repeated crystal growth experiment at 40 °C and 75% RH was conducted. The crystal growth study was then performed at 76 and 80% RH. The temperature was kept constant at 25 °C. Since the rate of crystallization is highly dependent on the environmental conditions, the experiments were stopped when the phase equilibrium was reached and no further crystal growth was detected. To determine the rate constants of crystal growth, the normalized crystal content was fitted by the Avrami phase transfer kinetics.19−21 2.6. Drug Release. Dissolution studies were performed using a USP apparatus 2 setup (Vision Elite 8 combined with a Maximizer “AutoPlus”, Hanson Research, Chatsworth, CA, USA) in 900 mL of medium with either 1.442% sodium dodecyl sulfate in water, a paddle speed of 75 rpm, and a bath temperature of 37 °C or 1.442% cetyltrimethylammonium bromide in 0.05 M phosphate buffer with 25 rpm and at 30 °C. The tablets were transferred into Japanese pharmacopeia sinkers to prevent the samples from floating. During the dissolution experiment, samples were taken after 15, 30, 60, 90, 120, 240, 480, 760, and 1440 min and filtered through a 10 μm cannula filter (UHMW-PE, Hanson Co., Dissolution Accessories, Oosterhout, The Netherlands). The drug release during the dissolution was quantified by HPLC (Waters 2695, Waters, Waters Corp., Milford, MA, USA), as described in our earlier work.16

Table 1. Overview of Tg (Closed Pan), Content, Impurities, Water Content, and Crystallinity of the Investigated Formulationa

a

sample

extrudate

tablet

Tg (°C) content (% LC) sum of impurities (% LC) water content (%(w/w)) cAPI PLM cAPI XRPD

41 99.4 0.1 2.8 -

40 100.4 0.1 2.3 N/A -

Values for the extrudate and for the tablet are given as applicable.

was determined to be 100% label claim (LC). This value is in compliance with the range of 95−105% LC, the typically used assay specification. The sum of impurities in the stored samples is below 0.5% (w/w). Hence, no significant chemical degradation of the API has taken place. No crystallinity was observed in the samples by means of XRPD (see the Supporting Information). Although XRPD is the standard technique for crystallinity quantification, the LOD of this technique is commonly in the magnitude of 0.1% (w/w). In PLM imaging, single birefringent structures can be observed (see the Supporting Information). Also, with PLM, no crystallinity was observed in the investigated samples. An interference of impurities or residual crystallinity with the crystal growth in this formulation is therefore not anticipated. 3.2. Crystal Growth Investigation. The crystal growth in the fenofibrate formulation was investigated by TRS. As the nondestructive nature of TRS allows for the quantification of the crystallinity in one and the same solid dosage form over the entire course of the crystallization process, this technique is well suited for the production of ASD samples with a defined amount of crystalline content. Additionally the measurement time of TRS is much shorter than the storage period of the

3. RESULTS AND DISCUSSION Fenofibrate is a well-known model substance in ASD research and an example of a fast crystallizing BCS II compound. The stability of an ASD is influenced by two environmental factors: temperature and moisture. Earlier studies conducted with fenofibrate formulations have shown that the crystallization of this particular API is substantially accelerated under elevated relative humidity. Different techniques, like DSC, XRPD, and Raman spectroscopy, have been used in the literature to monitor the crystallization of an ASD.15,22,23 The nondestructive nature of TRS allows for the consecutive measurement of one and the same dosage form over time. Additionally, C


Article


Figure 2. PLS regression overview. Calibration data points are depicted in blue and validation data points in red. Plot (i) shows the convergence of the RMSE and explained variance with an increasing number of latent variables. The scores of factor 1 vs factor 3 are shown in (ii). In (iii) the correlation between the predicted and reference values is given. The ideal regression is depicted as a black line. The calibration and cross-validation results are depicted as blue and red lines, respectively (the lines are superimposed), and the confidence interval (p = 95%) is depicted as black dashed lines. A comparison of the pretreated spectrum of cAPI (black) and the loading of factor 1 (red), factor 2 (blue), and factor 3 (green) in the used ROI is shown in (iv).

samples. Hence, no influence of the measurement on the crystal growth rate is anticipated. To quantify crystallinity in the fenofibrate formulation on the basis of TRS spectra, a chemometric calibration model needs to be established to perform multivariate data analysis. This quantification method has already been successfully applied in our investigation of a fenofibrate formulation stored for 15 years.16 In brief, a calibration set spanning up to 15% (w/w) cAPI was prepared. Spiked powder mixtures with different amounts of crystalline API were pressed into round tablets (500 mg, 13 mm diameter), and the spectra of three independently prepared samples were measured for each spiking level. In order to reduce noise and normalize the data, the second derivative spectrum (Savitzky-Golay algorithm, 5 point window, thirdorder polynomial) was calculated, and unit vector normalization was performed. A partial least-squares (PLS) regression model was calculated in the spectral region of interest (ROI) between 82 and 309 cm−1. An overview of important model metrics of the calculated PLS is given in Figure 2. After the third factor, no significant increase in the explained variance is observed. Hence, a good fit of the model to the data is achieved by using three latent variables. The scores plot shows an excellent separation of the individual spiking levels between factors 1 and 3. The samples with a spiking level of 6 and 7.5% (w/w) cAPI are not well separated in the two-dimensional score plot but are in a threedimensional representation (not shown). The reference values for the calibration are obtained by weighing the amount of spiked crystalline API. The calibration model shows a linear correlation between the reference values of the sample set and the predicted values of the PLS regression. The slope of the regression line as well as the slope of the cross-validation line

are very close to one, and their offset is very close to zero. The samples are randomly distributed around the regression line. The correlation coefficient is 0.99 for the calibration and 0.98 for the cross validation. The confidence interval (p = 95%) shows a sufficient precision for the intended purpose. Three factors result in an optimal fit of the model. Two factors are expected to fit the crystalline API and the excipients. Indeed, the explained variance of the system is 98% after two factors. The third factor contributes only 1.2% of variance but significantly lowers the root-mean-square error (RMSE). This part of the explained variance of the model can be assigned to spectral contributions of interactions between the amorphous API and the polymer, that, of course, decrease with an increasing content of cAPI. The comparison of the loading of the three factors with the TRS signal of the cAPI in the ROI after spectral preprocessing shows several similar features. The correlation with the reference values as well as the spectral composition of the loadings indicates that we indeed quantify crystalline API in the samples. The model was then applied to follow the crystal growth of fenofibrate in the formulation. To show that it is possible to follow the growth of crystals, samples of the fenofibrate formulation were stored under environmental stress of 40 °C and 75% RH. The TRS spectra of the samples were measured every 24 h, and the crystalline content was determined by using the chemometric model described above. Note that the overall exposure due to the measurements outside of the climate chamber is less than 20 min each day. This time period has no significant impact on the crystallization behavior of the sample. Three samples were measured at each time point. As the amount of crystallization reached its final value and the crystal growth curve leveled out, the experiment was stopped. In order D


Article

Molecular Pharmaceutics to ensure the reproducibility of the crystal growth, the experiment was repeated with a fresh sample set. To quantify the rate constant, the crystal growth trace was fitted according to the Avrami equation for phase transition in solids:19−21 α(t ) = 1 − exp[−k(t − t0)n ]

(1)

where α(t) is the relative crystallinity over time, the fraction of crystallinity relative to the final amount of crystallization in the sample; k is the rate constant; t0 is an offset time; and n is the Avrami exponent related to the dimension of crystal growth. The crystal growth curve of the two feasibility sample sets of crystal-free round tablets stored at 40 °C and 75% RH is depicted in Figure 3. Nearly identical amounts of crystalline

Figure 4. The kinetic trace of tablets of the fenofibrate formulation stored at 25 °C and 80% RH (i) and 76% RH (ii).

Table 2. Levels of Crystallinity and Corresponding Storage Times under Different Conditions Figure 3. The kinetic trace of a crystal growth experiment in the investigated fenofibrate formulation at 40 °C and 75% RH (black, average of three individual samples) and the repetition experiment under the same conditions (red, average of three individual samples).

material are quantified at each time point in both experiments, and the same equilibrium concentration of crystalline API is reached. Similar to other fenofibrate formulations in the literature, the Avrami exponent was set to 3 for fitting the data points.22 Nearly identical rate constants are obtained in both experiments (see the Supporting Information) by fitting the kinetic traces with the Avrami equation (eq 1). Under the investigated environmental conditions, the crystal growth curve is quite steep and does not allow for a controlled growth of crystalline API with the desired precision. Since fenofibrate is quite prone to crystallize under elevated humidity levels, it is possible to lower the temperature of the crystal growth experiment while raising the humidity to achieve a better controllable time scale. A temperature of 25 °C, combined with humidity levels of 80 and 76% RH, was observed to be most suitable for this formulation and the desired levels of crystallinity. As the kinetic function is not simply transferable to a change in both humidity and temperature, the crystal growth experiment was performed once at each of the altered environmental conditions to determine the rate of crystal growth (Figure 4). Using these kinetic traces, it is now possible to develop a storage scheme for the production of tablets with different levels of crystallized API. The storage times and conditions to achieve certain levels of crystallinity are given in Table 2. As the samples with grown crystallinity are intended for dissolution testing, football-shaped tablets were used. These tablets are in the same weight range and possess the same dimensions as material commonly used for clinical testing. To ensure that the chemometric model can still be applied to the football-shaped

cAPI (%) (w/w)

cAPI (% LC)

t25/80 (days)

t25/76 (days)

1.5 3 6

10 20 40

4.5 5.5 6.5

5.5 8 12.5

tablets, the TRS spectra in the ROI of both tablet types were compared (see the Supporting Information). Since no differences in the spectra are observed above noise level, the model can directly be transferred to the football-shaped tablets. Another factor that could potentially result in a different behavior is moisture uptake. The crystal growth experiments are conducted at very high relative humidity; however, moisture uptake is so fast24 that, within the sampling frequency applied, no differences could be determined for these two tablet shapes. Interestingly, it took up to three times longer to grow the desired amount of crystalline content in the tablets during the production run compared to the experiments, where the kinetic trace was determined. Nevertheless the desired amount of crystalline content could be reached in any of the conducted experiments. The root cause of the delay in crystal growth was readily determined. The number of tablets used during the production run was increased by a factor of 10. This amount of tablets presents a significant mass of hygroscopic polymer, depleting the RH in the climate chamber at the start of the crystal growth experiment. Since the rate of crystal growth in the fenofibrate formulation is very sensitive to moisture, the crystallization process was slower in the production run. Samples that were exposed to a lower humidity, and hence a slower crystal growth, show a narrower distribution in the observed crystallinity concentrations. Variations in crystal growth rate induced by small sample inhomogeneities, like trace amounts of residual crystallinity, minor variations in local drug load, and the like, seem to be the origin for this variation in the individual sample populations. Those factors have a E


Article


Figure 5. USP App. II dissolution profiles of tablets with different amounts of spiked crystallinity (left) and recrystallized API (right) (black, 0% (w/ w) cAPI; red, 1.5% (w/w) cAPI; blue, 3% (w/w) cAPI; and green, 5% (w/w) cAPI). For the upper row, dissolution data were obtained from a SDSbased medium and for the lower row, from a CTAB solution.

Table 3. Comparison and Evaluation of the Dissolution Curves of Amorphous Tablets and Tablets with Spiked Crystallinity (SC) by f1 and f 2 Tests amorphous

amorphous

SDS

1.5% SC

3.0% SC

5.8% SC

CTAB

1.5% SC

3.0% SC

5.8% SC

f1 f2

5 65

15 42

25 31

f1 f2

6 70

14 51

66 18

and 3.0%-spiked tablets compared to the amorphous tablets were confirmed with f 2 values above 50 (Table 3), whereas the highest spiking level differs with an f 2 of 31. Although the cAPI content in the tablets used for replicates was slightly different, similar release profiles were measured for the same targeted grown amount compared to the spiked samples (Figure 5ii). Overall, the release behavior of the tablets with grown crystallinity resulted in even more shallow dissolution profiles compared to those of the spiked tablets, but also they had similar end points after 24 h of 100.7 ± 0.7% LC. Because of the differences in the shape and completeness, no similarity factors were calculated for tablets with grown crystallinity (Figure 5ii,iv).27 Due to the observed similarities in dissolution profiles of tablets with spiked cAPI obtained by the drug release method developed in 2002, a more discriminative dissolution method was developed (Figure 5iii,iv), utilizing lower paddling speed and a lower temperature. This approach enables not only better discrimination between the dissolution curves throughout most of the experiment but also direct determination of crystalline content in spiked tablets based on the total released API (% LC) amount after 720 min (Figure 5iii). It can be speculated that the lower paddling speed and temperature of this dissolution method result in a thicker diffusion layer around

higher impact on the amount of crystallized API when the rate of crystal growth is faster. Therefore, environmental conditions facilitating a slower growth process lead to more precise amounts of crystalline material in the produced tablets. In the end, a compromise has to be found between crystal growth precision and the invested time to grow the crystals for the API under investigation. 3.3. Dissolution Testing. As the dissolution behavior of an ASD is a key metric for its performance, the drug release of the football-shaped tablets with grown cAPI was compared to the release behavior of football-shaped tablets that contained spiked cAPI. First, tests were performed with a modified dissolution method used in the initial development of the fenofibrate formulation in 2002 (Figure 5i,ii).16 For the amorphous fenofibrate tablets, about 90% LC release was achieved within 90 min, and full release (102.3 ± 0.8% LC) was achieved after 120 min (Figure 5i, black line). Dissolution profiles of tablets with spiked cAPI content showed a flattened progression, which increased with additional crystalline content; however, all resulted in a similar end point. To compare the dissolution curves statistically, the similarity and difference factors were calculated. For dissolution curves to be similar, the difference factor f1 should be below 15 and the similarity factor f 2 above 50.25,26 The similarities in the dissolution profiles of the 1.5%F


Article


for biopharmaceutical studies on the impact of crystallinity on the bioavailability.

the tablets. The higher solution concentration in the immediate area around the sample would lead to a better pronunciation of effects impacting the dissolution behavior. Amorphous and spiked tablets show a typical dissolution curve progression with an initial steeper segment, which continuously flattens until a plateau is reached. No significant difference was measured in the release of the amorphous and 1.5%-spiked tablets ( f 2 = 65); however, samples with higher spiking levels had f 2 values below 50. In contrast, tablets with grown crystallinity showed an almost linear dissolution behavior over 24 h and reached about half of the dissolved amount only at 720 min compared to the spiked tablets (Figure 5iv). Depending on the parameters used for dissolution testing, it was possible to modulate and amplify differences in the dissolution rate of the spiked fenofibrate tablets. The impact of different forms of crystallinity was shown by using two non-QC dissolution approaches. In both, spiked tablets showed a similar dissolution behavior in terms of erosion and tablet disintegration compared to the amorphous tablets, as reflected by the dissolution curve progressions. In contrast, the dissolution profiles of tablets with grown crystallinity revealed a completely different release behavior. In previous studies, the impact of crystallinity in ASD formulations on dissolution behavior has been investigated, e.g., with capsules filled with differently spiked blends,28 dissolution of powder samples with grown crystallinity,29 and spiked tablets.30 As the different dissolution behaviors of powders and capsules may not reflect the behavior of tablets, the use of spiked tablets is a valuable approach to develop a discriminative dissolution method. However, differences in the dissolution behavior of tablets with spiked and grown crystalline content indicate additional effects, e.g., matrix alterations during storage. With standard dissolution methods, the variations in the release behavior can be minimal and may be overlooked. The systematic study of factors influencing the release of partially crystalline ASD formulation will be the focus of future studies. To gain better insights about how crystallinity in ASD formulations may impact their performance, samples which reflect the reality as best as possible are needed. As shown in our current results, the use of tablets with grown crystallinity is an invaluable tool to assess changes in the tablet matrix as well as API during storage.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.molpharmaceut.8b00043. (H)PLM images of extrudate and tablet samples; XRPD patterns of extrudate and tablet samples; figures of merit of the chemometric calibration; rate constants, maximum cAPI, and half-life of the different crystal growth experiments; and a comparison of the TRS spectra of round and football-shaped tablets (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Tel.: +49 621 589 2037; Fax: +49 621 589 62037; E-mail: [email protected]. ORCID

Frank Theil: 0000-0002-9396-3118 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Numerous AbbVie employees of NCE Analytical R&D and NCE Formulation Sciences were involved in the manufacturing of the samples as well as routine analytics. Their contribution to this work is highly appreciated. All authors are employees of AbbVie. The design, study conduct, and financial support for this research were provided by AbbVie. AbbVie participated in the interpretation of data, review, and approval of the publication.

■

REFERENCES

(1) Lipinski, C. A.; et al. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Delivery Rev. 2001, 46 (1−3), 3−26. (2) Lipinski, C. A. Drug-like properties and the causes of poor solubility and poor permeability. J. Pharmacol. Toxicol. Methods 2000, 44 (1), 235−49. (3) Leuner, C.; Dressman, J. Improving drug solubility for oral delivery using solid dispersions. Eur. J. Pharm. Biopharm. 2000, 50 (1), 47−60. (4) Chiou, W. L.; Riegelman, S. Pharmaceutical applications of solid dispersion systems. J. Pharm. Sci. 1971, 60 (9), 1281−302. (5) Rumondor, A. C.; Taylor, L. S. Effect of polymer hygroscopicity on the phase behavior of amorphous solid dispersions in the presence of moisture. Mol. Pharmaceutics 2010, 7 (2), 477−90. (6) Lakshman, J. P.; et al. Application of melt extrusion in the development of a physically and chemically stable high-energy amorphous solid dispersion of a poorly water-soluble drug. Mol. Pharmaceutics 2008, 5 (6), 994−1002. (7) Worku, Z. A.; Aarts, J.; Van den Mooter, G. Influence of compression forces on the structural stability of naproxen/PVP-VA 64 solid dispersions. Mol. Pharmaceutics 2014, 11 (4), 1102−8. (8) Kaminska, E.; et al. Studying the Impact of Modified Saccharides on the Molecular Dynamics and Crystallization Tendencies of Model API Nifedipine. Mol. Pharmaceutics 2015, 12 (8), 3007−19. (9) Brandl, M.; Frank, K. J.; Rosenblatt; Holig; Rosenberg; Magerlein; Fricker, G.; Westedt. The amorphous solid dispersion of the poorly soluble ABT-102 forms nano/microparticulate structures in aqueous medium: impact on solubility. Int. J. Nanomed. 2012, 7, 5757−5768.

4. CONCLUSION To lower the pill burden and enable patient compliance, ASDs with higher drug loads are very desirable. Such high drug load ASDs are approaching the physical limitations of stabilization as well as miscibility and are naturally accompanied by a risk of crystallization over their shelf life. With the right tools in hand, we are able to provide formulations and the respective controls to avoid crystallization even over decades, as previously shown.14,16 In this work, we followed the crystal growth of individual tablets at accelerated storage conditions. Recording the kinetic trace of crystal growth enabled us to produce samples with defined crystallized API content for biopharmaceutical testing. The impact of ASD crystallinity on tablet dissolution behavior was investigated in both tablets with spiked and grown crystallinity. Since the only boundary conditions for this approach are that crystallization occurs on an observable time scale and that crystallinity can be quantified in the solid dosage form, the methodology is transferable to ASD formulations of other APIs. The new method for targeted crystallization of amorphous tablets to simulate crystallized ASDs enables us, for the first time, to generate model systems G


Article

Molecular Pharmaceutics (10) Van den Mooter, G. The use of amorphous solid dispersions: A formulation strategy to overcome poor solubility and dissolution rate. Drug Discovery Today: Technol. 2012, 9 (2), e79−e85. (11) Ambike, A. A.; Mahadik, K. R.; Paradkar, A. Spray-dried amorphous solid dispersions of simvastatin, a low tg drug: in vitro and in vivo evaluations. Pharm. Res. 2005, 22 (6), 990−8. (12) Yu, L. Amorphous pharmaceutical solids: preparation, characterization and stabilization. Adv. Drug Delivery Rev. 2001, 48 (1), 27−42. (13) Zhou, D.; et al. Crystallization kinetics of amorphous nifedipine studied by model-fitting and model-free approaches. J. Pharm. Sci. 2003, 92 (9), 1779−92. (14) Theil, F.; et al. Frozen in Time: Kinetically Stabilized Amorphous Solid Dispersions of Nifedipine Stable after a Quarter Century of Storage. Mol. Pharmaceutics 2017, 14 (1), 183−192. (15) Greco, S.; et al. A practical method to predict physical stability of amorphous solid dispersions. Pharm. Res. 2012, 29 (10), 2792−805. (16) Theil, F.; et al. Extraordinary Long-Term-Stability in Kinetically Stabilized Amorphous Solid Dispersions of Fenofibrate. Mol. Pharmaceutics 2017, 14 (12), 4636−4647. (17) Lehmkemper, K.; et al. Long-Term Physical Stability of PVPand PVPVA-Amorphous Solid Dispersions. Mol. Pharmaceutics 2017, 14 (1), 157−171. (18) Baghel, S.; Cathcart, H.; O’Reilly, N. J. Polymeric Amorphous Solid Dispersions: A Review of Amorphization, Crystallization, Stabilization, Solid-State Characterization, and Aqueous Solubilization of Biopharmaceutical Classification System Class II Drugs. J. Pharm. Sci. 2016, 105 (9), 2527−44. (19) Avrami, M. Kinetics of Phase Change. I General Theory. J. Chem. Phys. 1939, 7, 1103−1114. (20) Avrami, M. Kinetics of phase change, II. Transformation-time relation for random distribution of nuclei. J. Chem. Phys. 1940, 8, 212− 224. (21) Avrami, M. Granulation, phase change, and microstructure, kinetics of phase change. III, general theory. J. Chem. Phys. 1941, 9, 177−184. (22) Feng, X.; et al. Evaluation of the recrystallization kinetics of hotmelt extruded polymeric solid dispersions using an improved Avrami equation. Drug Dev. Ind. Pharm. 2015, 41 (9), 1479−87. (23) Kestur, U. S.; et al. Influence of particle size on the crystallization kinetics of amorphous felodipine powders. Powder Technol. 2013, 236, 197−204. (24) Hiew, T. N.; et al. Effect of moisture sorption on the performance of crospovidone. Int. J. Pharm. 2016, 514 (1), 322−331. (25) O'Hara, T.; Dunne, A.; Kinahan, A.; Cunningham, S.; Stark, P.; Devane, J. Review of methodologies for the comparison of dissolution profile data. In Vitro-in Vivo Correlations 1997, 423, 167−171. (26) Shah, V. P.; et al. In vitro dissolution profile comparison Statistics and analysis of the similarity factor, f(2). Pharm. Res. 1998, 15 (6), 889−896. (27) Duan, J. Z.; Riviere, K.; Marroum, P. In vivo bioequivalence and in vitro similarity factor (f2) for dissolution profile comparisons of extended release formulations: how and when do they match? Pharm. Res. 2011, 28 (5), 1144−56. (28) Rahman, Z.; et al. Comparison of X-ray powder diffraction and solid-state nuclear magnetic resonance in estimating crystalline fraction of tacrolimus in sustained-release amorphous solid dispersion and development of discriminating dissolution method. J. Pharm. Sci. 2015, 104 (5), 1777−86. (29) Kanaujia, P.; et al. Investigating the effect of moisture protection on solid-state stability and dissolution of fenofibrate and ketoconazole solid dispersions using PXRD, HSDSC and Raman microscopy. Drug Dev. Ind. Pharm. 2011, 37 (9), 1026−1035. (30) Pack, B. W.; et al. Development of an in vivo-relevant drug product performance method for an amorphous solid dispersion. J. Pharm. Biomed. Anal. 2017, 142, 307−314.

H


Manufacturing Amorphous Solid Dispersions with a Tailored Amount

Recommend Documents