Impact of Polymers on the Crystallization and ... - ACS Publications


Sep 3, 2014 - A major limitation of these formulations stems from the propensity of amorphous solids to ... place. Crystallization in the matrix will ...
0 downloads 0 Views 8MB Size

Article pubs.acs.org/molecularpharmaceutics

Impact of Polymers on the Crystallization and Phase Transition Kinetics of Amorphous Nifedipine during Dissolution in Aqueous Media Shweta A. Raina,† David E. Alonzo,‡,§ Geoff G. Z. Zhang,‡ Yi Gao,‡ and Lynne S. Taylor*,† †

Department of Industrial and Physical Pharmacy, College of Pharmacy, Purdue University, West Lafayette, Indiana 47907, United States ‡ Drug Product Development, Research and Development, AbbVie Inc., North Chicago, Illinois 60064, United States ABSTRACT: The commercial and clinical success of amorphous solid dispersions (ASD) in overcoming the low bioavailability of poorly soluble molecules has generated momentum among pharmaceutical scientists to advance the fundamental understanding of these complex systems. A major limitation of these formulations stems from the propensity of amorphous solids to crystallize upon exposure to aqueous media. This study was specifically focused on developing analytical techniques to evaluate the impact of polymers on the crystallization behavior during dissolution, which is critical in designing effective amorphous formulations. In the study, the crystallization and polymorphic conversions of a model compound, nifedipine, were explored in the absence and presence of polyvinylpyrrolidone (PVP), hydroxypropylmethyl cellulose (HPMC), and HPMC-acetate succinate (HPMC-AS). A combination of analytical approaches including Raman spectroscopy, polarized light microscopy, and chemometric techniques such as multivariate curve resolution (MCR) were used to evaluate the kinetics of crystallization and polymorphic transitions as well as to identify the primary route of crystallization, i.e., whether crystallization took place in the dissolving solid matrix or from the supersaturated solutions generated during dissolution. Pure amorphous nifedipine, when exposed to aqueous media, was found to crystallize rapidly from the amorphous matrix, even when polymers were present in the dissolution medium. Matrix crystallization was avoided when amorphous solid dispersions were prepared, however, crystallization from the solution phase was rapid. MCR was found to be an excellent data processing technique to deconvolute the complex phase transition behavior of nifedipine. KEYWORDS: multivariate curve resolution, polymorphism, crystallization, polymers, supersaturated solutions, Raman spectroscopy, solution mediated phase transformation



(matrix crystallization) or the solid first undergoes rapid dissolution forming a supersaturated solution followed by crystallization from solution (solution crystallization). Additionally, some combination of these two events may also take place. Crystallization in the matrix will minimize the degree of supersaturation generated, and rapid crystallization from solution will limit the amount of time that the supersaturated solution persists. Both routes of crystallization will nullify any solubility advantage provided by the amorphous form. Crystallization from either route involves nucleation, i.e., formation of critical clusters or seed crystals followed by growth.13 Polymers can be utilized to delay either of these processes and thus kinetically stabilize supersaturated solutions.14−18 The extent and duration of supersaturation achieved for a given system will depend on the properties of the API, in

INTRODUCTION Amorphous solid dispersions (ASDs) are an important formulation design strategy for poorly water-soluble drugs.1,2 Amorphous systems give rise to supersaturated solutions upon dissolution that can lead to improved drug absorption and enhanced bioavailability relative to conventional formulations. Although many studies have probed the influence of different polymers and drug−polymer interactions3−8 on the solid state stability of amorphous solid dispersions9,10 and during storage under various conditions, there are far fewer studies examining the crystallization behavior of amorphous systems during dissolution.11,12 A mechanistic understanding of crystallization and the kinetics of phase transitions during dissolution is important in order to effectively implement an ASD formulation. Dissolution studies examining crystallization of amorphous solids, such as those conducted by Alonzo et al.,11 have established that, when an amorphous system comes into contact with aqueous media during dissolution, there are two potential routes of crystallization: either the solid crystallizes © 2014 American Chemical Society

Received: Revised: Accepted: Published: 3565

May 5, 2014 August 18, 2014 September 3, 2014 September 3, 2014 dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

particular its inherent crystallization tendency (slow crystallizers may remain supersaturated for longer periods of time as compared to their fast crystallizing counterparts),16 the chemistry of the polymer,19 drug−polymer interactions,20,21 drug−polymer miscibility, the drug−polymer ratio, and temperature. Adding to the complexity of the crystallization process, the first crystalline form produced from either the hydrated matrix or the supersaturated solution may be a metastable polymorph that subsequently converts to the stable, lower energy form.22 In turn, both the kinetics of phase transformations and the outcome may be influenced by the use of polymers.23−25 For example, Price et al.24 demonstrated the selective crystallization of two different polymorphs of acetaminophen and all six forms of ROY using polymer templating techniques whereas Gift et al.26 showed that certain polymers substantially delayed the conversion of the metastable anhydrous form of carbamazepine to the thermodynamically stable hydrate in aqueous suspensions. The importance of different polymorphic forms is well established whereby it is known that the crystal form impacts not only the solubility and physicochemical properties of the API but also the properties of the drug product and its safety and efficacy in vivo.27 For example, the thermodynamically stable polymorph of chloramphenicol palmitate is biologically inactive.28 Of the three polymorphic forms of sulfameter, form II exhibited 1.4 times higher bioavailability as compared to the more stable form III.29 In the case of sulphamethoxydiazine, form II is twice as soluble as form III but has an absorption profile 40 times superior.29 The polyene antibiotics mepartricin and nystatin exhibit different polymorphs which have shown differences in both bioavailability and toxicity profiles.30 In order to successfully evaluate the dissolution behavior of amorphous solids, it is clearly important to understand the impact of polymers on crystallization kinetics and pathways as well as any stabilization effects on polymorphic transitions since each of these factors will potentially impact bioavailability. Figure 1 shows a schematic of the possible crystallization pathways from an amorphous solid during dissolution suggested by Alonzo et al.11 and has been modified to include possible polymorphic transitions and potential polymer impact. Figure 1 illustrates that an amorphous solid exposed to aqueous media may undergo initial crystallization to a metastable form (form I): either in solid matrix or from supersaturated solution.

Subsequently, the stable polymorph (form II) grows at the expense of the metastable form (form I). If dissolution of the solid matrix is slow, then crystallization likely results in the solid matrix. If, however, dissolution of the matrix is rapid, the solid dissolves into solution forming a supersaturated solution and then crystallizes initially as form I with subsequent conversion to form II. The goals of the current study were to investigate the crystallization and phase transition behavior of amorphous nifedipine systems focusing on the kinetic aspects, and to investigate the influence of polymers on such transitions. Nifedipine, a relatively fast crystallizer,31,32 was selected as the model compound as it is known to exist in several different polymorphic forms, all of which have been extensively characterized,33−36 thus leading to the likelihood of complex phase behavior. Polarized light microscopy (PLM) and Raman spectroscopy were used as the principal analytical techniques to monitor crystallization. It is challenging to obtain quantitative kinetic information on complex phase transitions using conventional methods which involve building calibration curves. Calibration samples are powder blends containing different ratios of physical mixtures of amorphous and crystalline forms. Even for a compound that exhibits a single crystalline form, producing calibration samples is timeconsuming and challenging as it may involve isolating and stabilizing the amorphous form. This is further compounded for systems with multiple polymorphs where complex multicomponent physical mixture calibration samples at several different ratios are required. Multivariate analysis is increasingly being used to extract information from spectral data. Multivariate analyses have been widely utilized to evaluate solid-state mixtures of polymorphs, the evolution of polymorphic forms in small molecules and protein formulations as well as to monitor degradation kinetics of pharmaceutical drugs.31,37−41 However, such methods have not been extensively applied to monitor crystallization in aqueous media. In this study, principal component analysis (PCA) and multivariate curve resolution (MCR) approaches were utilized to quantify crystallization kinetics and polymorphic transitions when amorphous nifedipine solids were exposed to aqueous media. These methods allow us to bypass calibration sets and obtain quantitative information about the transformation kinetics directly from spectroscopic data.

Figure 1. Schematic illustrating the competition between dissolution and crystallization via the solid or solution state for amorphous systems.

Nifedipine was purchased from Euroasia (Mumbai, India). Polyvinylpyrrolidone (PVP) K29/32 was purchased from Sigma-Aldrich Co. (St. Louis, MO, USA), whereas hydroxypropylmethyl cellulose (HPMC) Pharmacoat grade 606 and hydroxypropylmethyl cellulose acetate succinate (HPMC-AS) MF grade were obtained from ShinEtsu (Shin-Etsu Chemical Co., Ltd., Tokyo, Japan). The dissolution media used in all experiments comprised 50 mM pH 6.8 phosphate buffer with or without predissolved polymer at a concentration of 1 mg/ mL. No solubility enhancement was observed for nifedipine at this polymer concentration.17 Ethyl alcohol (200 proof) was purchased from Pharmco Products, Inc., Brookfield, CT, USA, and dichloromethane (ChromAR) was purchased from Mallinckrodt Baker Inc., Phillipsburg, NJ, USA. Dri-Rite was obtained from The Dri-Rite Company, Chicago, IL. Molecular structures of the model compounds and polymers are shown in Figure 2.



3566

MATERIALS

dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

dispersions of nifedipine were placed on an aluminum sample holder, and the surface was smoothed and leveled with a glass slide. Polarized Light Microscopy (PLM). In conjunction with PXRD, cross-polarized light microscopy (PLM) using a Nikon Eclipse E600 Pol microscope (Nikon Co., Tokyo, Japan) was used to confirm the amorphous nature of samples prior to use. PLM was also used to obtain estimates of crystallization onset times, to discriminate between the two aqueous routes of crystallization (crystallization of the dissolving matrix or precipitation from solution), and to evaluate the crystallization kinetics. Samples were sprinkled onto a glass slide, and 2−3 drops of 50 mM phosphate buffer with and without predissolved polymer were added. Samples were then covered with a glass coverslip (L × W × D: 22 × 70 × 1.0 mm, Thermo Fischer Scientific, Waltham, MA). Images and videos of samples were captured using NIS-Elements version 2.3 software package (Nikon Co., Tokyo, Japan) using 10× and 40× magnification. Raman Spectroscopy. Crystallization onset and crystallization kinetics of slurries of amorphous solids and solid dispersions were monitored using RAMANRXN2 HYBRID analyzer (Kaiser Optical Systems, Inc., Ann Arbor, MI). Slurries of amorphous nifedipine consisted of approximately 500 mg of solid added to 2 mL of 50 mM phosphate buffer pH 6.8 with or without predissolved polymer at 25 and 37 °C. Slurries of ASDs comprised 1g of solid in 4 mL of 50 mM phosphate buffer pH 6.8 without predissolved polymer at 25 and 37 °C. Spectra were collected using a fiber optically coupled noncontact PhAT probe, every minute for the first 15 min, and thereafter every 5 min for up to 3 h. Qualitative spectral analyses were carried out using carbon to carbon double bond stretch (CC) and ester (C−C−O) region peaks to discriminate between amorphous (henceforth denoted as “g” for glass), alpha (α), beta (β), and gamma (γ) polymorphs of nifedipine. In the CC region, g, α, β, and γ polymorphs exhibit peaks centered at 1647, 1648, 1651, and 1644 cm−1, whereas in the C−C−O region, these forms exhibit peaks centered at 1214, 1224, 1216, and 1205 cm−1 respectively.36 For MCR, several peaks in the spectral region 1000−1800 cm−1 were used to discriminate between the different forms. For brevity, the Results section will only discuss the C−C−O region. The β polymorph was never observed in the course of our experiments. Spectra were collected using iC Raman (Version 3.0, Kaiser Optical Systems, Inc., Ann Arbor, MI). Data were evaluated with Grams/AI (Version 7.02, Thermo Galactic, part of Thermo Fischer Scientific, Waltham, MA). Multivariate Data Analysis. In order to generate kinetic profiles of polymorphic changes conventionally, physical mixtures of polymorphs are used in varying ratios. Since nifedipine slurries exhibited three forms, namely, g, α, and γ, and generating an adequate number of samples of stable physical mixtures of metastable forms is very tedious and challenging, multivariate data analysis was utilized to aid in kinetic analysis. The Unscrambler software (Camo, NJ, USA) was used to carry out principal component analysis (PCA) and multivariate curve resolution (MCR) in order to identify the polymorphic changes and to obtain a kinetic profile of phase transitions. Determining the number of components is crucial to the MCR technique. Either a priori knowledge is essential or PCA analysis to determine components is required. Raman peaks in the region from 1000 to 1800 cm−1 were selected for the analysis. For PCA, a standard normal variate analysis was

Figure 2. Molecular structures of model compounds and polymers.



METHODS Preparation of Amorphous Solids and Solid Dispersions. Nifedipine was made amorphous by heating the crystalline API to 10 °C above the melting temperature followed by rapid cooling in liquid nitrogen. Amorphous samples were cryomilled with a Spex model 6750 (SPEX CertiPrep LLC., Metuchen, NJ) cryomill for 45 s at 5 impacts per second for 3 cycles yielding particles in the size range of 1− 300 μm. Amorphous nifedipine was used immediately after preparation. Amorphous solid dispersions of nifedipine (50:50 w/w) were prepared using a solvent evaporation technique. Nifedipine and polymer at a 1:1 weight ratio were dissolved in a 1:1 v/v mixture of ethanol (200 proof) and dichloromethane. Dissolution was aided by the use of sonication for up to 30 s. Organic solvent was rapidly removed under vacuum using rotary evaporation. Solid dispersions were dried overnight under vacuum followed by milling in a Spex model 6750 (SPEX CertiPrep LLC., Metuchen, NJ) cryomill for 45 s at 5 impacts per second for 3 cycles, which led to particles of 1−300 μm. The amorphous material obtained was stored in desiccators containing Dri-Rite and refrigerated at 4 °C until use. Powder X-ray Diffraction (PXRD). To confirm the amorphous nature of melt quenched nifedipine and the amorphous solid dispersions, powder X-ray diffraction (PXRD) was carried out using a Shimadzu XRD-6000 diffractometer (Shimadzu Corporation, Kyoto, Japan) equipped with a Cu Kα source and set in Bragg−Brentano geometry between 5 and 35° 2θ at a scanning speed of 5°/min with a 0.04° step. The diffractometer was calibrated using a silicon reference standard (NIST 640c) at one degree intervals. Data analysis was carried out using Grams/AI (Version 7.02) Thermo Galactic). Reference samples of crystalline and amorphous nifedipine in addition to amorphous solid 3567

dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

Figure 3. Crystallization of neat amorphous nifedipine in the absence and presence of polymers. Micrographs (A) show that no crystallinity was present initially and (B−E) that matrix crystallization occurs within 15 min. Photomicrographs B−E show neat amorphous nifedipine exposed to pH 6.8 50 mM phosphate buffer, 25 °C at 15 min after exposure.

between measured data and their decomposition are collected in matrix E.

carried out and the data was mean centered. PCA enabled identification of similarities or patterns in multivariate data matrices (in our case wavenumbers at different time points). It reduces a large number of variables, i.e., wavenumbers, into a handful of orthogonal components or principal components which describe maximum variability in data sets. The NIPALS algorithm was used with cross validation and an initial guess of 7 components. Close to 96% of the variability in all data sets evaluated was explained by 3 principal components. MCR was carried out on data sets which were corrected with a baseline offset and 20 points of smoothing. MCR is a calibration free approach which is used to resolve mixtures by determining the number of constituents, the response profiles, and the concentration profiles, and only requires an initial estimate of number of components or forms present in addition to pure spectra of each of the forms as reference. Normalized pure spectra of each form were used. No direct reference spectrum was obtained for the metastable γ-polymorph. Instead, a time point at which only γ-polymorph was present in the sample was used as the reference spectrum. The absence of other forms was confirmed by cross validating with spectral information from the literature. MCR is affected by rotational ambiguities. When determining kinetic concentration profiles, these can be minimized by setting constraints such as nonnegative values and closure constraint for mass balance. Constraints were set on non-negative spectra, non-negative concentrations, and closure. In addition, the sensitivity of components was set to 74 and a maximum of 200 ALS iterations were carried out. MCR involves deconvoluting the data matrix D into two smaller matrices C and ST (see eq 1). C is the matrix of concentration profiles of each of the pure components whereas ST is the matrix of pure components. Lastly, E represents matrix of residual error which has the same dimensions as D. E arises as the decomposition of matrix D is not perfect, and differences

D = CST + E

(1)

The initial MCR model was refined by applying additional constraints on non-negativity, and closure since mass balance is expected. In addition to providing concentration profiles of each of the forms as a function of time, MCR provides a plot of the resolved component spectra (see Figure 4B). Dissolution Studies. Dissolution experiments for pure amorphous nifedipine and a 50:50 nifedipine:HPMC ASD were carried out using a pION μ-Diss Profiler (pION Inc.,Woburn, MA) equipped with a 6-channel fiber-optic probe system, heating blocks, and magnetic stirrers (500 rpm), at 37 °C, using the same medium as for the Raman experiments. This in situ probe approach was used to capture any transient increases in concentration during dissolution, however, the solids loading (5 mg/mL) was much lower than for the Raman experiments since a high particle density made it impossible to obtain absorbance data due to scattering effects.



RESULTS Crystallization and Polymorphic Transitions from Neat Amorphous Nifedipine. As previously stated, amorphous solids may crystallize via two major routes upon exposure to aqueous media, namely, the solid matrix or solution routes (see Figure 1). Polarized light microscopy (PLM, Figure 3A,B) revealed that neat amorphous nifedipine in the presence of buffer crystallizes rapidly, primarily through the matrix route. The presence of polymers such as PVP, HPMC, and HPMCAS predissolved in the buffer did not impact the onset of crystallization or the route of crystallization of neat amorphous nifedipine, as shown in Figure 3C−E. Thus, even in the presence of dissolved polymers, crystallization of the solid matrix was instantaneous. 3568

dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

polymorph, as indicated by the peak at 1205 cm−1. After 5 min, signs of an additional polymorph, the α-form, begin to appear based on the evolution of a peak at 1224 cm−1. The relative normalized peak intensities at 1205 and 1224 cm−1 provide a qualitative estimate of the amount of γ-polymorph versus αpolymorph present in the slurry as a function of time. As time progresses, the normalized peak intensity at 1224 cm−1 increases and that at 1205 cm−1 decreases, indicating that the stable α-polymorph grows at the expense of the metastable γpolymorph. At 40 min, the majority of the crystalline material exists as the α-polymorph. However, at the end of 3 h, small traces of the γ-polymorph persist. These results are also summarized in Table 1. Multivariate curve resolution revealed the evolution of nifedipine polymorphs as a function of time. In agreement with the peak height ratio analysis described above, MCR shows a rapid decrease in the amount of amorphous nifedipine and the concurrent rapid increase in the amount of the γ-form, with the crystallization transition reaching completion within about 10 min (Figure 5B). Shortly after, conversion of the γpolymorph to the α-polymorph occurs. At 37 °C, the onset of crystallization in a slurry of neat amorphous nifedipine was also immediate, and the γ-polymorph transformed more rapidly to the α-polymorph. A comparison of the crystallization and polymorphic conversion kinetics at the two temperatures, based on MCR evaluation, is shown in Figure 6. The impact of predissolved polymer (PVP, HPMC, or HPMC-AS) added to the phosphate buffer on the crystallization behavior is shown in Figures 7 and 8 and Table 1. Table 1 summarizes manual spectral analyses (using peak height ratio analysis) of the crystallization behavior of neat amorphous nifedipine in the absence and presence of polymers at 25 and 37 °C, confirming the results of the MCR analysis in Figures 7 and 8. Although the polymers did not impede the onset of crystallization at either 25 or 37 °C, with the amorphous material disappearing within a few minutes, the presence of polymers did appear to have a stabilizing effect on the γpolymorph, delaying its conversion to the stable, α-form. Concentration versus time profiles showing the evolution of each of the polymorphic forms are summarized based on MCR data in Figure 7 (formation of γ-polymorph at 25 and 37 °C) and Figure 8 (formation of α-polymorph at 25 and 37 °C). The relative effectiveness of each of the polymers in stabilizing the γpolymorph follows the rank order HPMCAS > HPMC > PVP whereby the trends were similar at each temperature, although the conversion kinetics between the polymorphic forms were found to be faster at 37 °C. Crystallization and Polymorphic Transition from Nifedipine Amorphous Solid Dispersions. Previous studies have shown that polymers such as PVP and HPMC can inhibit matrix crystallization of small molecules such as felodipine.11 Since polymers predissolved in solution did not appear to be able to inhibit the rapid matrix crystallization of amorphous nifedipine in our studies, it was interesting to investigate how amorphous nifedipine would behave when intimately mixed with a polymer at the molecular level, i.e., as an ASD. ASDs of nifedipine were formulated with PVP, HPMC, and HPMCAS at 50:50 drug−polymer ratios. Table 2 summarizes the crystallization behavior of nifedipine solid dispersions as obtained from manual spectral analysis. The time points in Table 2 indicate the time at which Raman peaks characteristic of each polymorph were first detected. MCR plots (Figures 9A and 9B) summarize the polymorphic transition kinetics at 25

To explore the crystallization kinetics and polymorphic transitions more quantitatively, Raman spectroscopy was carried out. The Raman shifts most commonly used to discriminate between the glass (g), α, and γ polymorphs of nifedipine are peaks in the C−C−O stretching region (1214, 1224, and 1205 cm−1 for the three forms respectively) and C C/CO stretching region (1647, 1680, and 1644 cm−1 for g, α, and γ forms respectively).36 In this study, several peaks in the spectral region 1000−1800 cm−1 were used for form identification. However, for brevity the discussion will be limited to the C−C−O region. The β-polymorph was not detected in our studies. Figure 4A shows reference Raman spectra from the C−C−O region, i.e., 1190−1235 cm−1, from which it is apparent that the different forms can be readily discriminated.

Figure 4. Raman spectra of different forms of nifedipine at 25 °C: (A) reference spectra from manual measurements of pure forms and (B) resolved component spectra obtained from MCR analysis of spectra obtained during the transformation of amorphous nifedipine to different crystalline forms following exposure to buffer.

MCR was carried out to calculate relative concentration profiles from Raman spectra. A plot of the extracted component spectra is shown in Figure 4B, where it can be seen that the spectra are in excellent agreement with the reference spectra shown in Figure 4A. The spectra in Figure 5A show that, immediately after addition of buffer to neat amorphous nifedipine, the slurry has spectral signatures from both the amorphous form (g), as indicated by the peak at 1214 cm−1, and the metastable γ3569

dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

Figure 5. (A) Raman spectra obtained from of a slurry of amorphous nifedipine in buffer at pH 6.8, 25 °C, at various time points following exposure to the solvent. (B) MCR processed data showing the decrease in the amount of amorphous nifedipine (glass) as a function of time with the concurrent increase in the amount of the alpha form, followed by polymorphic transformation of the gamma form to the alpha polymorph.

Table 1. Summary of Crystallization Kinetics of Neat Amorphous Nifedipine in the Absence and Presence of Polymers at 25 and 37 °C Using Peak Height Ratio Analysisa

a

Glass (g) or other polymorphic forms depicted in red indicate that, at the given time, the majority of the system exists as that particular form. Small amounts of the form in black may still be present The data are in good agreement with the MCR analysis shown in Figures 7 and 8.

and 37 °C. Although none of the three polymers delayed the onset of crystallization, the cellulosic polymers HPMC and HPMCAS appeared to be superior at impeding the conversion of γ- to α-polymorph as compared to the PVP system. Further, as previously observed, both polymorphs formed at a faster rate at 37 °C than 25 °C. MCR plots of crystallization kinetics (Figures 9A and 9B) and results from manual spectral analyses (Table 2) are in good agreement. Surprisingly, both the conversion rate of amorphous to gamma and that of the gamma polymorph to the stable alpha

form were much faster for the ASDs than for neat amorphous nifedipine (Figures 7A and 8B). To follow up on this somewhat perplexing observation, polarized light microscopy images were obtained and a dissolution study was carried out. Images for nifedipine:PVP dispersions are shown in Figure 10, where it is apparent that the dissolving solid dispersion particles do not crystallize. However, small plate-shaped crystals rapidly evolve from the solution phase. These images strongly suggest that the route of crystallization has changed from the matrix to the solution phase. Nifedipine in the amorphous solid dispersion 3570

dx.doi.org/10.1021/mp500333v | Mol. Pharmaceutics 2014, 11, 3565−3576

Molecular Pharmaceutics

Article

Figure 6. Concentration versus time profiles for the γ-polymorph (A) and the α-polymorph (B) formed following exposure of amorphous nifedipine to buffer at 25 and 37 °C.

Figure 7. Evolution kinetics of the γ-polymorph from aqueous slurries of neat amorphous nifedipine in the presence of polymers at 25 °C (A) and 37 °C (B).

Figure 8. Evolution kinetics of the α-polymorph from aqueous slurries of neat amorphous nifedipine in the presence of polymers at 25 °C (A) and 37 °C (B).

α-polymorph. The dissolution data is thus consistent with the rapid crystallization of neat amorphous nifedipine in the matrix, as described above. Lastly, to investigate if the lack of crystallization in the ASD matrix was due to polymers effectively inhibiting matrix crystallization or primarily on account of the rapid dissolution of the solid dispersion, the pH of the dissolution medium was changed in order to reduce the dissolution rate. HPMCAS is an ionizable polymer and is not very soluble at pH 2.42 Thus, the dissolution rate at this pH is substantially reduced and the matrix does not dissolve. The insoluble matrix can then be evaluated for crystallization as a function of time. Images are shown in Figure 12, where it is apparent that the matrix does not crystallize. However, clearly some drug is released from the

thus appears to be rapidly dissolving into solution, forming supersaturated solutions, and then crystallizing from solution. By comparing Figures 3 and 10 it is apparent that the crystalline material produced by the solution route leads to smaller particles/agglomerates (