Published online 22 June 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI /jps.22242

Transcription

1 PHARMACEUTICAL TECHNOLOGY Physicochemical Properties of the Amorphous Drug, Cast Films, and Spray Dried Powders to Predict Formulation Probability of Success for Solid Dispersions: Etravirine ILSE WEUTS, 1 FREDERIC VAN DYCKE, 1 JODY VOORSPOELS, 1 STEVE DE CORT, 1 SIGRID STOKBROEKX, 1 RUUD LEEMANS, 1 MARCUS E. BREWSTER, 1 DAWEI XU, 2 BRIGITTE SEGMULLER, 2 YA TSZ A. TURNER, 3 CLIVE J. ROBERTS, 3 MARTYN C. DAVIES, 3 SHENG QI, 4 DUNCAN Q.M. CRAIG, 4 MIKE READING 4 1 Chemical and Pharmaceutical Development, Johnson & Johnson, Beerse, Belgium 2 Analytical Development, Johnson & Johnson Pharmaceutical R&D, LLC, Raritan, New Jersey Laboratory of Biophysics and Surface Analysis, School of Pharmacy, University of Nottingham, University Park, Nottingham, NG7 2RD, United Kingdom 4 School of Chemical Sciences and Pharmacy, University of East Anglia, Norwich, United Kingdom Received 31 January 2010; revised 23 April 2010; accepted 26 April 2010 Published online 22 June 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI /jps ABSTRACT: Solid dispersion technology represents an enabling approach to formulate poorly water-soluble drugs. While providing for a potentially increased oral bioavailability secondary to an increased drug dissolution rate, amorphous dispersions can be limited by their physical stability. The ability to assess formulation risk in this regard early in development programs can not only help in guiding development strategies but can also point to critical design elements in the configuration of the dosage form. Based on experience with a recently approved solid dispersion-based product, Intelence 1 (etravirine), a three part strategy is suggested to predict early formulate-ability of these systems. The components include an assessment of the amorphous form, a study of binary drug/carrier cast films and the evaluation of a powder of the drug and polymer processed in a manner relevant to the intended final dosage form. A variety of thermoanalytical, spectroscopic, and spectrophotometric approaches were applied to study the prepared materials. The data suggest a correlation between the glass forming ability and stability of the amorphous drug and the nature of the final formulation. Cast films can provide early information on miscibility and stabilization and assessment of processed powders can help define requirements and identify issues with potential final formulations. ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 100: , 2011 Keywords: solid dispersion; characterization; amorphous; etravirine; HPMC INTRODUCTION The development of new pharmaceutical agents is often confounded by their poor water solubility which impacts a number of derivative properties including poor dissolution rate. 1,2 Inadequate solubility or dissolution rate can significantly reduce both the rate and extent of drug absorption deleteriously affecting oral bioavailability. Several factors rooted Correspondence to: Ilse Weuts (Telephone: þ ; Fax: ; iweuts@its.jnj.com) Journal of Pharmaceutical Sciences, Vol. 100, (2011) ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association in contemporary drug pipelines are thought to contribute to this overall decrease in drugability including the manner in which drugs are discovered and the chemically allowed space of the drug target. As reviewed by Lipinski, 1,3 high throughput screening has tended to select for compounds with lower water-solubilities, higher lipophilicities and higher molecular weights. Furthermore, many drug targets have structure-activity relationship (SAR) requirements that do not overlap with properties known to provide for good oral bioavailability resulting in poor develop-ability and high attrition. 4,5 While expanding the chemical space of the target is the purview of medicinal chemistry, pharmaceutical scientists can 260 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

2 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 261 contribute to enlarging the oral drug bioavailability space through the use of technologies that increase apparent solubility and dissolution rate. A number of approaches are available for increasing the formulate-ability of BCS Class II compounds including but not limited to complexation, micronization or nanonization, use of cosolvents and structured lipid systems. 6 8 Solid amorphous dispersions provide a unique opportunity in this regard While theoretically of benefit in any number of situations, the physical instability of solid amorphous dispersions often conspires to limit their usefulness. Early assessments of factors likely to impact on the applicability of a dispersion based approach such as stability, are important components in any development program centered around such a formulation concept. We suggest herein a three pronged approach for assessing dispersion fitness using the recently approved drug, Intelence 1 (etravirine) as an illustrative example. Etravirine (Fig. 1) is a novel second generation nonnucleoside reverse transcriptase inhibitor (NNRTI) which acts by molecularly blocking the viral reverse transcriptase enzyme. It achieves this by preventing the enzyme from converting its genetic material (RNA) into proviral DNA and thus preventing incorporation of the viral genome into the human host cell. 12,13 The physicochemical properties of etravirine are challenging with regard to dosage form configuration as indicated in Table 1. A variety of technologies were screened to optimize bioavailability, however most did not add significant value. For example, oral dosing of a nanosuspension resulted in negligible blood levels in dog. The best enhancement of GI uptake was provided by systems based on an amorphous drug substance and specifically, amorphous dispersions of the compound in a glassy carrier. 9 11,14,15 These initial screening elements were the inputs for a development program to fabricate a viable formulation for etravirine. In devising this program, factors thought to contribute to, or detract from success were posited. The philosophy adopted included (1) a study of the pure amorphous drug phase, (2) an evaluation of cast films of the drug and a glassy polymer (HPMC), and (3) an assessment Figure 1. Chemical structure of etravirine. Table 1. Physicochemical and Solubility Properties of Etravirine Property of powders prepared using scaled-down processes simulating those likely to apply to the finished dosage form. Thus, this communication examines the properties of amorphous etravirine, 16 the characteristics of solvent cast films containing the drug and HPMC and assessments of the physical state of the drug and its miscibility with the polymer as a function of drug/ polymer ratio in spray dried powders. 15,17 20 These assessments required a number of appropriate and complementary analytical methods such as powder X-ray diffraction (XRD), infrared spectroscopy (IR), differential scanning calorimetry (DSC), modulated temperature differential scanning calorimetry (M-DSC) and solid state nuclear magnetic resonance (SS-NMR). Imaging techniques such as field emission gun-environmental scanning electron microscopy (FEG-ESEM) and atomic force microscopy (AFM) allowed the spatial assessment of pharmaceutical formulations with resolution up to the nanometer scale. EXPERIMENTAL Value Molecular weight (g/mol) 435 Molecular formula C 20 H 15 BrN 6 O Melting point 2608C (dec.) log p >5 pk a (base) <3 S (water) 1 mg/ml S (0.1 N HCl) 1 mg/ml S (PEG400) 74 mg/ml Materials Etravirine (Fig. 1) was obtained from Janssen Pharmaceutica (Beerse, Belgium). The hydroxypropylmethyl cellulose (HPMC) was 2910 grade with a viscosity of 5 mpa s and was purchased from Dow Chemical (Plaquemine, LA). Amorphous etravirine was prepared by cryomilling the crystalline drug substance for 3 h at 1968C. Other methods such as spray drying, freeze drying, and ball milling were unsuccessful in generating stable amorphous etravirine (data not shown). Stability was assessed using FT-IR as described in the literature. 16 Film Casting A screening study found two useful solvent mixtures including dichloromethane (DCM)-ethanol (80:20) and various compositions of dimethylformamide (DMF)-methanol (80:20, 50:50, and 40:60). Drug and polymer (HPMC) were dissolved in the solvent mixture of interest and the solutions were filtered and DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

3 262 WEUTS ET AL. cast onto a Teflon plate (15 cm 15 cm). The solvent was evaporated in a vacuum oven at between 85 and 1258C for several days until a constant weight was obtained. Films were then subjected to M-DSC for determination of T g and the enthalpy of relaxation as well as XRD for assessing crystallinity. Spray-Dried Powders Solid dispersions of various compositions were prepared by spray drying on a laboratory scale instrument (NIRO atomizer mobile minor, Soeberg, Denmark). Etravirine and HPMC were dissolved in an organic solvent mixture (dichloromethane (DCM)/ethanol), after which the solvent was removed by spray drying. A dual fluid nozzle with a diameter of 1 mm was applied. Solutions were sprayed using a feed rate of 3 4 kg/h, an atomization pressure of 0.8 bars, an inlet temperature of 1108C and an outlet temperature of 708C. These parameters were maintained for the different dispersions. However, the solvent composition and the concentration of etravirine and HPMC in the feed solutions varied (generating changes in solvent evaporation rate and solution viscosity). Powders with the following etravirine/hpmc ratios (w/w) were prepared: 1:12, 1:9, 1:6, 1:5, 1:3, 1:2, 1:1, 1:0.5, 1:0.3, 1:0.1, and 0:1. Physical mixtures were prepared by gently mixing amorphous, cryomilled etravirine and spray dried HPMC using a pestle and mortar. Powders with the following etravirine/hpmc ratios (w/w) were prepared: 1:3, 1:1, and 1:0.5. Modulated Temperature Differential Scanning Calorimetry (M-DSC) The cast films or powder samples were analyzed using a TGA Q-1000 MDSC (TA Instruments, Crawley, UK) with a heating rate of 28C/min, an amplitude of C and a period of 60 s. Standard aluminum DSC pans (TA Instruments) were used without crimping. Differential Scanning Calorimetry (DSC) The assessment of the T g of the amorphous etravirine could most easily be performed using thermal analysis without temperature modulation. The powder was heated in the same pan type as for the M-DSC measurements but higher heating rates were applied (from 10 up to 1008C/min). Powder X-Ray Diffraction Analysis (P-XRD) P-XRD-analyses were performed on a Philips X Pert PRO MPD diffractometer PW3050/60 with a PW3040 generator (PANalytical, Almelo, The Netherlands). The instrument is equipped with a Cu LFF X-ray tube PW3373/10. The cast films or powders were applied to a zero background holder and placed on a spinner stage with a spinner revolution time of 1 s. For quantitation of crystalline content including the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011 preparation of calibration curves, a sample holder with a cavity of 16 mm diameter was implemented. This methodology provided for improved robustness, increased sensitivity and was less susceptible to preferred orientation effects than the zero background holder. Bragg-Brentano geometry was applied using monochromatic Cu Ka-radiation and soller slits (0.04 rad) for the incident and diffracted beam path. The instrument was operated at a voltage of 45 kv and a current of 40 ma. The scan range was from 3 to 508 2u with a step size of and a counting time of 60 or 100 s per step. Crystallinity was estimated by constructing a standard curve in which spray-dried powders of HPMC and crystalline etravirine were mixed with amorphous powders providing a range from 0% to 100% crystalline content. The peak area associated with the diffraction peak between 8 and 108 2u (see Fig. 3b) was assessed as a function of the percent crystalline content. There was a good linear response (r 0.99) and low variability (samples of the calibration curve were measured in triplicate) at crystalline contents up to approximately 30% or 40%. At higher crystalline compositions, the linear fit of the data decreased and variability increased. The limit of detection was approximately 2% and the lower limit of quantitation was approximately 6%. Fourier-Transform Infra-Red Spectroscopy (FT-IR) FT-IR spectra were measured using a Nicolet Nexus 670 FT-IR spectrometer (Thermo Fisher Scientific, Waltham, MA) equipped with micro attenuated total reflectance (microatr). Thirty two scans were taken for each sample in the spectral range from 400 to 4000 cm 1 with a spectral resolution of 1 cm 1. Fourier-Transform Raman Spectroscopy (FT-Raman) The powders were transferred to a glass capillary cell and spectra were recorded on a Nicolet FT-Raman module (Thermo Fisher Scientific). 128 or 256 scans were taken for each sample in the spectral range from 100 to 4000 cm 1 with a spectral resolution of 4cm 1. Solid State-Nuclear Magnetic Resonance Spectroscopy (SS-NMR) The number of samples that were analyzed using this technique was limited to the following: pure components (Etravirine (both amorphous and crystalline) and HPMC), solid dispersions (ratios 1:3, 1:1, and 1:0.5) and physical mixtures of the same ratios. Solid-state carbon CP/MAS NMR data were collected using a Bruker Avance 400 MHz widebore NMR spectrometer equipped with a 4 mm MAS probe (Bruker BioSpin Corporation, Billerica, MA), with the Etravirine formulation samples spun at 5000 Hz and a contact time of 3.5 ms at 258C. Measurements DOI /jps

4 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 263 of amorphous Etravirine and of its physical mixtures with HPMC were performed at low temperature ( 508C), due to the thermal instability of the amorphous state in these samples. Proton spinlattice relaxation times (T 1 and T 1r ) were measured as well. Field Emission Gun Environmental Scanning Electron Microscopy (FEG-ESEM) The powders were sprinkled onto double sided carbon attached to a standard SEM platform, and any excess powder on the surface was removed by a stream of nitrogen. The uncoated samples were studied using a Philips FEI XL30 ESEM-FEG under a stream of nitrogen gas. The vapor pressure was kept between 1.8 and 1.9 Torr to minimize interference. A low accelerating voltage of 10 kv at a working distance of 10 mm and a beam current of 3 4 na were employed. Atomic Force Microscopy (AFM) A small amount of powder was sprinkled on to the surface of a thin layer of araldite glue. After 10 min, excess powder was removed under a gentle stream of nitrogen. Topographic height and phase images were simultaneously obtained using tapping mode AFM (TM-AFM) with a Nanoscope IIIa Multimode AFM system equipped with either an E or J scanner (Veeco Instruments TM, Santa Barabra, CA) in air. NP tips on 125 mm silicon cantilevers (Tap300, Budget Sensor TM, Sophia, Bulgaria) with a resonant frequency between 200 and 400 khz were used. Scan rates between 1 and 2 Hz were used to help minimize tip-sample interactions. RESULTS Amorphous Drug Substance The availability and study of pure amorphous etravirine would add value to the characterization of solid dispersions by not only providing a standard to assess drug polymer mixtures but also as a tool to gauge the nature of the formulation challenge and suggest the best development options to address these issues A number of approaches were attempted to generate the amorphous material. Based on the thermal lability some of these were not appropriate for this compound such as those related to melting and melt quenching. Spray-drying, ball milling, freeze-drying, rapid precipitation and rapid solvent evaporation (using a variety of conditions including flash evaporation using hyper-dsc) all provided for material containing significant crystalline content. The amorphous material was ultimately prepared by cryomilling (3 h at 1968C). 16 The amorphous nature of the material was confirmed by spectroscopic (FT- IR, FT-Raman, SS-NMR) and thermal techniques (DSC, M-DSC) as well as powder XRD. The IR- and Raman-spectra clearly show a distinct spectrum for the forms of interest in which the amorphous material was characterized by the absence of sharp absorption bands which are typical for crystalline systems. The FT-IR spectra of cryomilled etravirine and the crystalline polymorphic Forms I and II are shown in Figure 2. Three regions where the different forms demonstrate the greatest divergence in spectral characteristics have been highlighted. Similarly, peak broadening is observed in SS-NMR- 13 C-spectra (data not shown). XRD-analysis showed the absence of discrete diffraction peaks. A halo, typical for materials that lack long range order, is observed. The diffraction pattern of amorphous etravirine and two of its crystalline forms are shown in an overlay in Figure 3a and b. M-DSC-analysis indicates a glass transition (T g )at998c however the T g observed in the reversible heat flow signal is accompanied by recrystallization of the drug (exothermic event in the irreversible and total heat flow signal). In order to ensure that the phenomenon ascribed to the T g observed in the reversible heat flow is a true T g and not an artifact caused by a spurious deconvolution of the total heat flow associated with the reversible and nonreversible components, nonmodulated DSC measurements were also performed. High heating rates can inhibit kinetic phenomena such as cold crystallization or delay them to higher temperatures. 27 Figure 4 shows the DSC-curves that were recorded at different heating rates. A T g around 998C was observed, supporting the assumption that the phenomenon that was observed in the reversible heat flow curve of the M-DSC, is the glass transition. The anomalously high DC p at 1008C/min may be associated with better differentiation of the glass transition and cold crystallization events which may overlap at lower scanning rate. The generated thermoanalytical data can shed light on the stability of the amorphous phase (Tab. 2). The T g /T m ratio for a variety of pharmaceutical glass formers falls in a range of and can provide a general sense of glass fragility and the tendency to crystallize This value is difficult to generate for etravirine based on the fact that compound melting occurs coincident with degradation. To estimate the melting point, DSC experiments using high to very high scanning rates were employed. These data suggested a T g /T m of 0.69 which could suggest a fragile system. 28 The stability of the generated form was next assessed. Anecdotal information found that the amorphous material began to crystallize when stored in open conditions at ambient temperature in a few days. This was confirmed by a controlled assessment using FT-IR to assess spectral changes as a function of time. Even when the material was stored at 308C below its glass transition temperature, significant and rapid form conversion occurred. 16 DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

5 264 WEUTS ET AL. Figure 2. FT-IR spectra (from top to bottom): cryomilled etravirine, etravirine polymorph II, and polymorph I. Drug HPMC Cast Films Screening studies were performed to select an optimal common solvent from which drug polymer (HPMC) films could be cast. 31 DCM-ethanol allowed for useful films but only to a drug loading of 20% w/w. Above this drug polymer ratio, a crystalline component in the glass was evident upon casting. Improved results were found with DMF-methanol which allowed amorphous films with up to a 70% w/w drug loading to be prepared. This finding was interesting in that even though the evaporation rate of this solvent systems is likely to be lower than that of the DCM ethanol mixture (based on the differences in solvent vapor pressures) which would be expected to aid drug crystallization, the solubility of the components in the DMF methanol solvent was higher, reducing the extent of supersaturation and the crystallization driving force. Films were therefore prepared using this solvent at different drug and HPMC compositions. 31 The T g varied as a function of drug polymer ratio suggesting that a higher polymer content increased T g (drug/polymer ratio 1:1; T g ¼ 1068C, 1:2 ratio; T g ¼ 1118C and 1:3 ratio; T g ¼ 1148C). The cast films were then evaluated using M-DSC to assess the enthalpies of relaxation. The extent of relaxation at three drug to polymer ratios namely 1:1, 1:2, and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY :3 was assessed (Tab. 3). As a function of annealing temperature (i.e., storage temperature below T g ) and time, the extent of relaxation varied consistent with theory. Furthermore, the rank order suggested that the higher the polymer content, the lower the extent of relaxation and the greater the thermodynamic stability. While devitrification appears to be disfavored at higher polymer ratios, other kinetic factors can drive drug crystallization. 32 To assess these, XRD was applied to the cast films and the approximate extent of crystallization assessed (using a standard curve). The conditions examined included storage temperatures above the T g such that crystallization disconnected from relaxation could be studied. Within the limitations of this method (see the Experimental Section), a higher polymer content resulted in a longer induction period when the films were stored at T g (Tab. 4) and on a lower crystalline content and crystallization rate at temperature above their T g. These data point to the dramatic stabilizing effect of the polymer when the drug is made amorphous by film casting in HPMC in comparison to the cryomilled (amorphous) API. Processed Drug HPMC Powders Drug polymer powders prepared using the processing technique of interest were assessed. Powders DOI /jps

6 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 265 Figure 3. (a) Overlay of XRD-patterns of amorphous dispersions with different etravirine/hpmc ratio (from top to bottom: etravirine/hpmc 1:3, 1:1, 1:0.5, amorphous etravirine, spray-dried HPMC). (b) Overlay of XRD-patterns of partially amorphous dispersions with different etravirine/hpmc ratio (from top to bottom: etravirine/hpmc 1:0.3, 1:0.1, etravirine polymorph II, polymorph I). were prepared by spray-drying the drug at different ratios (etravirine/hpmc) specifically: 1:12, 1:9, 1:6, 1:5, 1:3, 1:2, 1:1, 1:0.5, 1:0.3, 1:0.1, and 0:1 w/w (spray dried HPMC). In addition, physical mixtures of the drug and HPMC were also generated at ratios of 1:3, 1:1, and 1:0.5. Figure 5 shows an overlay of the reversing heat flow curves of etravirine/hpmc dispersions with varying drug/polymer ratio. All the dispersions exhibit a single T g that falls in between the T g s of the two dispersed components (998C for etravirine and 1478C for HPMC), which indicates that all the dispersions are at least partially amorphous. Importantly, these data do not exclude the presence of some crystalline etravirine in the samples. In any case, the T g was found to increase with decreasing drug/polymer ratio which implies that the composition of the amorphous phase is dependent on the drug/polymer ratio. In Figure 6, the measured T g -values are plotted as a function of drug concentration. The T g changes with composition especially in polymer rich dispersions. In the drugrich dispersions (greater than 1:1 drug/polymer) T g s in the range of the pure amorphous phase were recorded and the changes in T g as a function of DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

7 266 WEUTS ET AL. Figure 4. DSC thermograms assessing T g at different heating rates for cryomilled (amorphous) etravirine. composition were less dramatic. Figure 6 also gives the theoretical T g -values calculated using the Gordon Taylor equation. 33 This relationship can be used to estimate the T g in binary mixtures; however, this relationship gives accurate predictions only in the case of ideal mixing, that is, when two conditions are being met: both components are mixed at the molecular level (no phase separation) and mixing is not accompanied by volume changes, that is, homoand heteromolecular forces are of equal strength. The Gordon Taylor equation is given as: T gmix ¼ w 1T g1 þ Kw 2 T g2 w 1 þ Kw 2 where w 1 and w 2 are the weight fractions of etravirine and HPMC and T g1 and T g2 are their glass transition temperatures. T g,mix is the glass transition temperature of the solid dispersion. K is a constant which can Table 2. Thermal Properties Related to Amorphous Etravirine and Itraconazole Property Etravirine Itraconazole Melting point 537 K a K Glass transition temp. (T g ) 373 K K Heat capacity (C p )att g 0.23 J/g K 0.43 J/g K Enthalpy of fusion (DH fus ) 100 J/g a 84.5 J/g T m /T g Activation energy of structural 905 kj/mol relaxation at T g Fragility parameter (m) 103 a Melts with decomposition. be approximated by the Simha Boyer rule: 34 K ¼ r 1T g1 r 2 T g2 where r 1 and r 2 are the densities of the API and the polymer. The T g values and true densities for etravirine and HPMC are 373 K (Tab. 2), g/cm 3 (density of crystalline etravirine) and 413 K and g/cm 3 for HPMC. The density of amorphous etravirine was estimated as 0.95 that of the crystalline density based on the literature, resulting in a value of g/cm Table 3. The Extent of Relaxation (%) as a Function of Annealing Temperature and Storage Time in Cast Films of Etravirine and HPMC Time (h) Drug/Polymer 1:1 (%) Drug/Polymer 1:2 (%) Drug/Polymer 1:3 (%) T g 25 K T g 35 K T g 45 K JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011 DOI /jps

8 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 267 Table 4. The Extent of Crystallization (%) as a Function of Temperature and Storage Time in Cast Films of Etravirine and HPMC Time (h) Drug/Polymer 1:1 (%) Drug/Polymer 1:2 (%) Drug/Polymer 1:3 (%) T g þ25 K T g þ10 K 3 4 a <LOD <LOD a <LOD a T g 3 <LOD <LOD <LOD 8 <LOD <LOD <LOD 24 3 a <LOD <LOD 96 6 <LOD <LOD a 3 a a 3 a a 5 a LOD, limit of detection. a Estimated value (between the limit of detection and the limit of quantitation). It is clear from Figure 6 that the experimental T g s deviate from the calculated values. Figure 7 shows an overlay of the total heat flow curves of the spray-dried dispersions with different drug/polymer ratios. They illustrate that once the temperature is above the T g, etravirine crystallizes causing the observed exothermic event in the DSC-curves. The data indicate that the barrier to crystallization increases as the polymer load increases, that is, crystallization takes place at higher temperature. Melting of etravirine takes place above 2008C (temperature region not included in the graph) and is accompanied by decomposition. Hence heats of fusion had to be estimated by rapid scanning techniques. As a consequence, the presence of crystalline etravirine in the spray dried dispersions cannot be accurately assessed by comparing the heat of fusion in the dispersions to the crystallization enthalpy of the drug. The DSC-analyses, therefore, suggest that the etravirine/hpmc materials consist of one amorphous phase (only one T g is measured for the dispersions). Thermal analysis does not exclude the possibility that in addition to this amorphous phase, crystalline etravirine is present in the dispersions. In the spray-dried powder, even at the lowest assessed ratio (etravirine/hpmc ¼ 1:0.1), a clear, single T g is observed, suggesting that most material is in the amorphous state. In order to assess whether small amounts of crystalline drug substance were present, other analytical techniques were Figure 5. Overlay of reversing heat flow curves for etravirine/hpmc spray dried formulations of different composition. DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

9 268 WEUTS ET AL. Figure 6. T g temperature-concentration profiles of spray-dried powders of etravirine and HPMC (~ experimentally determined values; & values calculated using the Gordon Taylor equation). applied including powder XRD, FT-IR/FT-Raman, and SS-NMR. Figure 3a and b shows an XRD overlay of the diffraction patterns of different dispersions. Only the spray-dried powders with a drug/polymer ratio of 1:0.1 and 1:0.3 exhibit discernable diffraction patterns that are characteristic for crystalline etravirine (polymorph Form I). Even in these cases, however, the XRD-patterns show only minor peaks that are superimposed on a halo that originates from the amorphous material present in the sample. These XRD-analyses confirm the M-DSC-results, that is, the 1:0.3 and 1:0.1 formulations are predominantly amorphous. FT-IR and FT-Raman analyses do not indicate the presence of crystalline etravirine in any of the dispersions. This suggests that these spectroscopic techniques are less sensitive than XRD in detecting low levels of crystalline drug. SS-NMR analysis was only performed on samples containing an etravirine/hpmc ratio of 1:0.5, 1:1, and 1:3. The CP-MAS- 13 C-measurements supported the conclusion that the drug was largely amorphous in these samples. 36,37 Specifically, proton T 1 and T 1r - values were measured both for the drug and the polymer (Tab. 5). The same was done for the physical mixtures (PM) of amorphous (cryomilled) etravirine and HPMC. In the solid dispersions, etravirine and HPMC have similar T 1 and T 1r -values, pointing to effective spin diffusion of the components and miscibility. The physical mixtures on the other hand, show a considerable difference in the 1 H T 1r -values of drug and polymer, confirming the heterogeneous nature of these materials. Visualization techniques were also applied to better characterize the spray-dried dispersions. Microscopic images generated with FEG-ESEM (Fig. 8), indicate that the dispersions consist of a mixture of spherical structures which were found to be hollow after focused ion beam scanning electron microscopy (FIB-SEM) assessment. The size of these spheres and the extent to which they have collapsed is Figure 7. Overlay of the total heat flow curves of etravirine/hpmc spray dried powders at different compositions. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011 DOI /jps

10 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 269 Table 5. Relaxation Parameters of the Etravirine/HPMC Spray Dried Powders (SD) and Physical Mixtures (PM) Measured with SS-NMR Sample Etravirine/HPMC dependent on the etravirine/hpmc ratio and the manufacturing process parameters. As mentioned in the Experimental Section, solvent composition, concentration of API and HPMC (and linked to this viscosity) were variables that were not the same for all prepared spray dried materials. It is well known for instance that spray drying under wet conditions (slow evaporation of the solvent) results in more collapsed spheres. Temperature settings were kept constant for the preparation of the samples but the solvent composition was not. Having more or less of the most volatile solvent present in this case would relate to more or less wet conditions. In addition, the feed solutions that were used had different viscosities which impacts particle size. The more viscous the feed, the larger the particles that are formed. No particles with other morphologies were observed. In Figure 9, AFM images of etravirine powders, ranging from high to low drug loading, are shown both as 3D topography (odd numbered images) and phase-based renderings (even numbered images). The topography and phase images of the relatively low polymer-loaded spray-dried powders, that is, etravirine/hpmc 1:0.1 (i and ii) and 1:0.3 (iii and iv), show distinct phase-separated areas and evidence of some ordered crystalline material. The AFManalyses confirm the predominantly amorphous nature of these samples which was established with the techniques outlined above (M-DSC, FT-Raman, FT-IR, and XRD). The AFM images also confirm the presence of some crystalline drug as was suggested by XRD. The spray-dried powder (drug/hpmc) 1:0.5 (v and vi) show some phase-separated domains but no crystalline material. Powders containing 1:1 to 1:9 (drug/hpmc) (images vii xvi) all show homogenous topography with no indication of phaseseparation or crystalline material. DISCUSSION 1 H T 1 (s) 1 H T 1r (ms) Etravirine HPMC Etravirine HPMC SD 1/ SD 1/ SD 1/ PM 1/ PM 1/ PM 1/ The optimal development strategy for solid dispersions, in general, and the etravirine dispersion, in particular, is best predicated using a number of early, efficient, and predictive screening experiments. These include, but are not limited to, those which probe physical stability. To configure a suitable roadmap, factors thought to contribute to, or detract from stability need to be identified, assays need to be developed and the aggregate information has to be interpreted to provide an estimation of technical risk as well as advice on the best excipients and processing tools. The approach suggested herein includes three evaluation elements including: (1) an assessment of the pure amorphous phase, (2) an evaluation of binary glasses cast from an appropriate solvent and (3) the study of processed powders using down-scaled production methods. The first component of this roadmap assesses the enablement likely required to generate a pharmaceutically relevant dispersion. The properties of the amorphous glass and its impact on the final dispersion depend on a number of factors including the domain size in the dispersion and miscibility, the possible molecular interaction of the amorphous phase with the glassy carrier and the mixing T g systems and storage conditions. 9 11,29 The greater the instability of the glass, the more important these stabilizing factors become. The data on etravirine presented here, as well as that presented in the literature 16 indicate that it is difficult to form a glass and once the glass is formed it is unstable even given its apparent high T g and taking the T g 50 rule into account. 38,39 It is interesting to compare these properties with those of another drug that has been marketed as a solid dispersion or solution. Itraconazole, for example, can be easily converted to an amorphous phase using any number of techniques including melt quenching, spray drying, lyophilization rapid precipitation, milling, etc. (Tab. 2). Furthermore, even with a measured T g of 598C, the prepared amorphous phase is stable to crystallization, even in open conditions for months to years The stability of the amorphous form may be suggested by its T g /T m value of 0.76 (compared to 0.69 for etravarine), its high energy of activation for glassy to super-cooled liquid transition and by analysis using Williams Watts formalism where storage of the material at 308C below the T g provide for a t value of 245 days and when stored at 408C below T g a t value of 80 years was calculated. 41,42 This stability is thought to be related to the ability of itraconazole to form a chiral nematic mesophase (i.e., a two dimensionally structured glass) and this finding directly impacts dispersion robustness and processing possibility. 40,43,44 The practical consequence of the relative stability of the itraconazole amorphous phase is that a number of formulation concepts can increase oral bioavailability relative to the crystalline drug including those based on complexation, bead coating, melt extrusion, spray-drying DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

11 270 WEUTS ET AL. Figure 8. FEG-ESEM images of etravirine/hpmc 1:1 (a and b), etravirine/hpmc 1:3 (c and d), etravirine/hpmc 1:6 (e and f) and broken particles to demonstrate the hollow nature of the spheres (g and h). and even tableting the amorphous form per se By contrast, the properties of etravirine strongly suggest that the number of formulation options may be limited and that stabilization of the amorphous phase will be paramount to achieving success. The data generated in this first part of the suggested roadmap JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011 therefore triggers a number of additional early screens and assessments. The examination of a cast film provides insight into general miscibility and compatibility and provides an easily testable, standardized matrix for assessing thermal properties, enthalpic relaxation DOI /jps

12 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 271 Figure 9. AFM-images of formulations etravirine/hpmc with ratios 1:0.1 (i and ii), 1:0.3 (iii and iv), 1:0.5 (v and vi), 1:1 (vii and viii), 1:2 (ix and x), 1:3 (xi and xii), 1:5 (xiii and xiv), and 1:9 (xv and xvi). (thermodynamic factors) and crystallization (kinetic factors)/phase separation within certain limits. 31 Studies on etravirine and HPMC films pointed to the stabilizing effects of a HMPC-based dispersion on the drug glass and that increasing the polymer component increased the relative stability. This was manifested by an increasing T g as the drug/polymer ratio decreases, meaning that molecular mobility is reduced at a given storage temperature. The measured values (see the Results Section) deviate from the theoretical T g s calculated using the Gordon Taylor equation. The deviation could have several causes including limited miscibility or phase separation. 51,52 Having considered this, the mixing T g s, being intermediate between the T g s of HPMC and amorphous etravirine, suggested a drug/ polymer dispersion. Also, the fact that a discernable T g for HPMC was not present suggests that a pure polymer phase was absent. Another cause for this deviation may be the fact that the zero volume change on mixing assumption is not met. 51,52 Violation from the free volume theory may be driven be differences in molecular interactions such that the homo- and heteromolecular associations are different. A derivative parameter that can give insight into dispersion stability is the extent of relaxation or the degree to which the glass approaches its theoretical equilibrium condition (DH 1 ¼ lim (t!1) DH which can be estimated by DH 1 ¼ (T g T)DC p ). 29,53,54 Relaxation is assessed by annealing samples at different storage temperatures below the T g and then assessing the change in relaxation enthalpy. Ordinarily the enthalpy of relaxation increases with increasing storage time and decreases with lower temperature in a nonlinear fashion. The Williams Watts relationship is a two parameter relationship (F t ¼ 1 (DH/ DH 1 ) ¼ exp (t/t) b ) that can suggest the time course of the relaxation process allowing a prediction of thermodynamic stability through the derived t (mean relaxation time constant) value. 11,55 Enthalpy of relaxation measurements for etravirine in cast HPMC films suggested that there was a dramatic effect of polymer content and annealing temperature. In the 1:3 drug/polymer glasses, the extent of relaxation was 2%, 6%, and 22% after 24 days of storage at 45, 35, and 258C below T g, respectively. For the same conditions in the 1:1 drug/polymer system, these values were 30%, 55% and 61% demonstrating DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

13 272 WEUTS ET AL. the stabilizing effect of the polymer, the reduced molecular mobility and expected long t value. However, it should be noted that thermodynamic processes are only part of the picture and building a dispersion only based on relaxation stability may be inappropriate as the kinetic components may be more important depending on the particular system of interest. 32 Kinetically, factors that impede crystallization include vitrification of the amorphous drug in the glassy polymer preventing the formation of a crystalline lattice. In order to investigate this aspect of dispersion stability, the rate and extent of crystallization were assessed. This was conducted at temperatures at, or above the T g, a domain where relaxation does not, theoretically, occur. These are accelerated studies and the kinetic component of crystallization from the dispersion may or may not mimic that at ambient temperature, however these approaches do provide a way to assess the effect of polymer content in the cast films on induction time and crystallization extent. These parameters were estimated by preparing a standard curve for XRD analysis. The data show that the 1:3 drug/polymer systems crystallized more slowly than the 1:2 which in turn was more stable than the 1:1 drug/polymer dispersions. This is associated with the extent of crystallization at the end of the experiments (200 h) for the 1:3 systems measured at T g, T g þ 10 and T g þ 25 of 5% (96 h indication time), 12% (8 h indication time) and 31% (crystalline detected at the first sampling point), respectively. The comparable data for the 1:1 drug/polymer system was 7% (8 h induction time), 25% and 43% (crystalline detected at the first sampling point in the latter two cases) at T g, T g þ 10 and T g þ 25, respectively. These experiments, based on the cast films, suggest that both thermodynamic (by a reduction in the enthalpy of relaxation with time) and kinetic (by reduced crystallization at and above the T g ) contributions could be responsible for the stabilization of the formed dispersions. 32 It is important to mention that while useful in screening mode, the cast films can contain large amounts of residual solvent significantly affecting the mixing T g s and related properties. This suggests that these data should be used qualitatively to suggest trends in drug stabilization. The advantage of this decisionmaking element includes the fact that the approach is fast, automatable in a 96-well format and information-rich, however they should not be overinterpreted. To better assess the role of processing and to provide a clearer view of the stability of the possible formulation components, down-scaled techniques are suggested to prepare powders for analytical scrutiny. This last component of the evaluation roadmap is important given that many amorphous dispersions, even when containing the same excipients at the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011 same levels, often manifest different stability and biopharmaceutical properties based on how they were generated. 56,57 Examples of down-scaled processes include the use of film forming or mini-fluid bed systems for coated bead systems, mini-extrusions for melt extrusion and the use of mini-spray-driers for this production technique. Similar to the approach suggested in the case of the film study, data derived from spray-dried powders indicated an increasing T g with increasing polymer content which was associated with higher stability and lower crystalline content (Figs. 5 and 6). As with the films, the data deviated from Gordon Taylor behavior suggesting factors associated with homo- and heteromolecular forces may be in play. Interestingly, the measured T g s were comparable for the 1:1 spray-dried drug/polymer mixture (1078C vs. 1068C for the cast film) but tended to be progressively higher in value at the 1:2 (1138C (powder) vs. 1118C (film)) and 1:3 compositions (1188C (powder) vs. 1148C (film)) lending some credence to the proposition that the processing technique could play a role in dispersion structure and performance. Factors that may be germane in this regard may include the efficiency of solvent evaporation or the nature of drug polymer mixing, that is, domain size and drug distribution. CONCLUSIONS Solid dispersions can allow for important drugs to reach the market by overcoming biopharmaceutical limitations imposed by Noyes Whitney and Fick s First Law effects. Poor physical stability is one factor that can limit the use of these potentially enabling systems and whether this property impacts dispersion applicability should be screened as early in the development cycle as possible to increase efficiency and probability of formulation success. Three inputs are suggested in making these aforementioned predictions including the stability and characteristics of the pure amorphous phase, the behavior of the amorphous drug when cast in films containing a glassy carrier and when dispersed in a polymer using a relevant process approach. The data suggest that etravirine is a poor glass former and the amorphous phase is unstable even given its high T g. By contrast, itraconazole manifests a much lower T g and yet forms a particularly stable amorphous phase. While highly unstable as such, cast drug/hpmc films resulted in a dramatic stabilization of amorphous etravirine with high polymer contents yielding progressively more stable dispersions. This information may point to the fact that multiple dispersion formulation are possible with itraconazole while only the spray-dried dosage form provided good oral DOI /jps

14 PREDICTION OF FORMULATION PROBABILITY OF SUCCESS FOR SOLID DISPERSIONS 273 bioavailability for etravirine (when compared to other amorphous systems). The generation and study of spray-dried powders allowed for a useful understanding of the formulation elements such that the ultimate formulation could be optimally designed and produced. REFERENCES 1. Lipinski CA Drug-like properties and the causes of poor solubility and poor permeability. J Pharmacol Toxicol Methods 44: Lipinski CA, Lombardo F, Dominy BW, Feeney PJ Experimental and computatorial approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev 46: Lipinski CA Avoiding investment in doomed drugs. Curr Drug Discov 1: Lipinski C, Hopkins A Navigating chemical space for biology and medicine. Nature 432: Lipinski CA The anti-intellectual effect of intellectual property. Curr Opin Chem Biol 10: Liu R Water insoluble drug formulation. Interpharm Press. Denver, CO. 7. Dressman J, Butler J, Hempenstall J, Reppas C The BCS: Where do we go from here? Pharm Tech 25: Brewster M, Loftsson T Cyclodextrins as pharmaceutical solubilizers. Adv Drug Deliv Rev 59: Leuner C, Dressman J Improving drug solubility for oral delivery using solid dispersion. Eur J Pharm Biopharm 50: Serajuddin AT Solid dispersion of poorly water-soluble drugs. Early promises, subsequent problems and recent breakthroughs. J Pharm Sci 88: Hancock BC Disordered drug delivery: Destiny, dynamics and the Deborah number. J Pharm Pharmacol 54: Andries K, Azijn H, Thielemans T, Ludovici D, Kukla M, Heeres J, Janssen P, De Corte B, Vingerhoets J, Pauwels R, de Béthune MP TMC125, a novel next-generation nonnucleoside reverse transcriptase inhibitor active against nonnucleoside reverse tgranscriptase inhibitor-resistant human immunodeficiency virus type 1. Antimicrob Agents Chemother 48: Vingerhoets J, Azijn H, Fransen E, De Baere I, Smeulders L, Jochmans D, Andries K, Pauwels R, de Béthune MP TMC125 displays a high genetic barrier to the development of resistance: Evidence from in vitro selection experiments. J Virol 79: Habib MJ Fundamentals of solid dispersions. In: Habib MJ, editor. Pharmaceutical solid dispersion technology. Boca Raton, FL: CRC Press. pp Craig DQM The mechanisms of drug release from solid dispersions in water-soluble polymers. Int J Pharm 231: Qi S, Weuts I, De Cort S, Stokbroekx S, Leemans R, Reading M, Belton P, Craig DQM An investigation into the crystallization behavior of an amorphous cryomilled pharmaceutical material above and below the glass transition. J Pharm Sci 99: Shah B, Kakumanu VK, Bansal AK Analytical techniques for quantification of amorphous/crystalline phases in pharmaceutical solids. J Pharm Sci 95: Six K, Murphy J, Weuts I, Craig DQM, Verreck G, Peeters J, Brewster M, Van den Mooter G Identification of phase separation in solid dispersions of Itraconazole and Eudragit 1 E100 using microthermal analysis. Pharm Res 20: Bauer-Brandl A Polymorphic transition of cimetidine during manufacture of solid dosage forms. Int J Pharm 140: Shi HG, Farber L, Michaels JN, Dickey A, Thompson KC, Shekular SD, Hurter PN, Reynolds SD, Kaufman MJ Characterization of crystalline drug nanoparticles using atomic force microscopy and complementary techniques. Pharm Res 20: Zhang J, Ebbens S, Chen X, Jin Z, Luk S, Madden C, Patel N, Roberts CJ Determination of the surface free energy of crystalline and amorphous lactose by atomic force microscopy adhesion measurements. Pharm Res 23: Ward S, Perkins M, Zhang J, Roberts CJ, Madden CE, Luk SY, Patel N, Ebbens SJ Identifying and mapping surface amorphous domains. Pharm Res 22: Price R, Young PM Visualization of the crystallisation of lactose from the amorphous state. J Pharm Sci 93: Hancock BC, Zografi G Characteristics and significance of the amorphous state in pharmaceutical systems. J Pharm Sci 86: Yu L Amorphous pharmaceutical solids: Preparation, characterization and stabilization. Adv Drug Deliv Rev 48: Ping G Amorphous pharmaceutical solids: Characterization, stabilization and development of marketable formulations of poorly water soluble drugs with improved oral absorption. Mol Pharm 5: Mathot VBF New routes for thermal analysis and calorimetry as applied to polymeric systems. J Therm Anal Cal 64: Chawla G, Bansal AK A comparative assessment of solubility advantage from glassy and crystalline forms of a water-insoluble drug. Eur J Pharm Sci 32: Crowley KJ, Zografi G The use of thermal methods for predicting glass-former fragility. Thermochim Acta 380: Wang L, Angell CA, Richert R Fragility and thermodynamics in nonpolymeric glass-forming liquids. J Chem Phys 125: Verreck G, Six K, Van den Mooter G, Baert L, Peeters J, Brewster ME Characterization of solid dispersions of itraconazole and hydroxypropylmethylcellulose prepared by melt extrusion Part 1. Int J Pharm 251: Bhugra C, Pikal MJ Role of thermodynamic, molecular and kinetic factors in crystallization from the amorphous state. J Pharm Sci 97: Gordon M, Taylor JS Ideal copolymers and the secondorder transitions of synthetic rubbers. I. Non-crystalline copolymers. J Appl Chem 2: Simha R, Boyer RF On a general relation involving the glass temperature and coefficients of expansion of polymers. J Chem Phys 37: Van den Mooter G, Van den Brande J, Augustijns P, Kinget R Glass forming properties of benzodiazepines and co-evaporate systems with poly(hydroxyethyl methacrylate). J Therm Anal Calor 57: Lubach J, Xu D, Segmuller B, Munson EJ Investigation of the effects of pharmaceutical processing upon solid-state NMR relaxation times and implications to solid-state stability. J Pharm Sci 96: Geppi M, Mollica G, Borsacchi S, Veracini CA Solid-state NMR studies for pharmaceutical systems. Appl Spectrosc Rev 43: Hancock BC, Shamlin SL, Zografi G Molecular mobility of glassy pharmaceutical solids. Pharm Res 12: DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011