RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Matrix-Assisted Cocrystallization: The Simultaneous Production and Formulation of Pharmaceutical Cocrystals by Hot-Melt Extrusion KEVIN BOKSA, ANDREW OTTE, RODOLFO PINAL Department of Industrial and Physical Pharmacy, Purdue University, West Lafayette, Indiana Received 1 February 2014; accepted 25 March 2014 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23983 ABSTRACT: A novel method for the simultaneous production and formulation of pharmaceutical cocrystals, matrix-assisted cocrystallization (MAC), is presented. Hot-melt extrusion (HME) is used to create cocrystals by coprocessing the drug and coformer in the presence of a matrix material. Carbamazepine (CBZ), nicotinamide (NCT), and Soluplus were used as a model drug, coformer, and matrix, respectively. The MAC product containing 80:20 (w/w) cocrystal:matrix was characterized by differential scanning calorimetry, Fourier transform infrared spectroscopy, and powder X-ray diffraction. A partial least squares (PLS) regression model was developed for quantifying the efficiency of cocrystal formation. The MAC product was estimated to be 78% (w/w) cocrystal (theoretical 80%), with approximately 0.3% mixture of free (unreacted) CBZ and NCT, and 21.6% Soluplus (theoretical 20%) with the PLS model. A physical mixture (PM) of a reference cocrystal (RCC), prepared by precipitation from solution, and Soluplus resulted in faster dissolution relative to the pure RCC. However, the MAC product with the exact same composition resulted in considerably faster dissolution and higher maximum concentration (∼five-fold) than those of the PM. The MAC product consists of high-quality cocrystals embedded in a matrix. The processing aspect of MAC plays a major role on the faster dissolution observed. The MAC approach offers a scalable process, suitable for the continuous manufacturing and C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci formulation of pharmaceutical cocrystals. Keywords: cocrystals; matrix; formulation; extrusion; continuous manufacturing; dissolution; supersaturation; crystal engineering; polymers; crystallization; solid dispersions R
INTRODUCTION Among the different formulation strategies for addressing the low systemic exposure of poorly soluble compounds administered by the oral route, cocrystallization is a promising approach that remains relatively underexplored in terms of formulation improvement. Pharmaceutical cocrystals, defined as solid-state stoichiometric molecular complexes of a drug molecule and a complementary coformer molecule, have demonstrated the ability to improve the solubility and physical stability of drugs.1–3 Nevertheless, their applicability in the pharmaceutical industry remains limited due in good measure to the lack of a suitable large-scale production method. Established cocrystal production methods rely on the generation of an intermediate phase with increased molecular mobility, such as the liquid or amorphous state, from which the components cocrystallize.4–6 Such methods include cocrystallization from a eutectic melt of the cocrystal components, cocrystallization via neat or liquid-assisted grinding (where the liquid acts as a catalyst), and cocrystallization by precipitation from solution. While each of these methods is capable of generating cocrystals, each also poses some limitations. Thermal methods involving melting require elevated temperatures that can compromise the integrity of thermolabile compounds. Besides being energetically inefficient, mechanical energy input methods like grinding can induce amorphization and their effectiveness tends to be limited without the use of a catalytic solvent. Finally, in methods
Correspondence to: Rodolfo Pinal (Telephone: +765-496-6247; Fax: +765-4946545; E-mail: [email protected]) Journal of Pharmaceutical Sciences
C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association
based on precipitation from solution, the location of the system in the phase diagram is critical to the cocrystallization process and to the attributes of the cocrystals produced,7 requiring both dynamic and precise control of supersaturation levels of the concentration of the component concentrations as cocrystallization takes place. This type of coordinated control can be difficult to maintain.6,8–10 A scalable production method for pharmaceutical cocrystals that draws on the advantages of existing methods, such as the control of stoichiometry offered by the thermal and energy input methods, along with the high crystal quality offered by the solution based methods, would be of significant valuable to pharmaceutical scientists during preformulation and formulation development. An equally important aspect for realizing the full potential of pharmaceutical cocrystals is the availability of a robust, scalable production method that effectively addresses the drawbacks (such as the potential chemical and physical stability issues mentioned above) associated with the cocrystal formation methods. In response to the need for an improved process, matrixassisted cocrystallization (MAC) is introduced here as a novel method of cocrystal production. Medina et al.11 have shown that hot-melt extrusion (HME) can be used as an effective cogrinding process to create cocrystals. The MAC approach utilizes HME to produce cocrystals embedded in a formulation matrix. Briefly, equimolar quantities of drug and coformer are mixed with a matrix material in the solid state prior to feeding the mixture into a hot-melt extruder. The extruder is set at a temperature where only the matrix material is made fluid, either by softening or by actual melting. Cocrystallization occurs during the extrusion process, induced by the intimate mixing and grinding of the components in the softened (or liquefied) matrix. The cocrystal particles formed this way become embedded in
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
1
2
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
the matrix material, which solidifies upon exiting the extruder. The end product of the MAC process, termed MAC product in this report, is a matrix-embedded cocrystal with the level of controlled stoichiometry offered by both thermal and mechanical cocrystal production methods. The MAC product is a formulated cocrystal where the matrix material plays a double role. During the HME process, the matrix plays role analogous to that of a catalytic solvent. In the final extrudate, the matrix is a functional component of the formulation. In MAC, the use of the melted/softened matrix phase promotes intimate mixing, while reducing excessive shear stresses produced when subjecting dry solid materials to extrusion. The reduced shear stress of MAC abates potential crystal damage. Producing cocrystal particles in functional matrices makes the physicochemical properties of the matrix as important as those of the cocrystal itself. Therefore, selection of the matrix material can be exploited to impart additional functionality into the formulated cocrystal, such as improved flowability, compaction, drug release kinetics, and so on. Thus, MAC serves as a method of simultaneous production and formulation of pharmaceutical cocrystals. In this report, we examine the ability of the MAC process to produce high-quality cocrystals. We also present the results of characterization studies comparing the physical attributes of the MAC product with those of a reference cocrystal (RCC) material, that is, the same cocrystal obtained by the careful implementation of the solution cocrystallization method. The study also includes a comparison of the in vitro dissolution performance between the MAC product and the RCC.
MATERIALS AND METHODS Carbamazepine (CBZ) and nicotinamide (NCT) were used as the drug and coformer, respectively, for the model cocrystal system of this study. Figure 1 shows the structures of the CBZ–NCT cocrystal and that of crystalline CBZ. CBZ is a BCS class II compound, with dissolution rate-limited absorption. One important aspect is that the anhydrous form of crystalline CBZ rapidly transforms into the dihydrate form during dissolution in aqueous media. This poses a problem because the transformation to the dihydrate results in a physical form that is less soluble, with considerably slower dissolution than the original anhydrous form. NCT was used as coformer for this study because it has been reported to form a cocrystal with CBZ whose solubility is 152-fold greater than that of the CBZ dihydrate.1 Soluplus was utilized as the matrix agent for cocrystal formation because of its relatively low glass transition temperature (∼72◦ C), allowing for a convenient processing window during extrusion. In addition, Soluplus has been shown to solubilize drugs and prevent their precipitation from supersaturated solutions.12,13 CBZ (CAS #298–46–4, Fig. 1b) was purchased from Alfa Aesar (Ward Hill, MA); NCT (CAS #98–92–0) was purchased from Sigma–Aldrich (St. Louis, MO); and Soluplus (CAS #402932–23–4) was obtained from BASF (Florham Park, NJ). All materials were used as received. In a typical MAC experiment, 3.2 g of a 1:1 molar ratio (66:34 mass ratio) CBZ–NCT mixture and 0.8 g Soluplus were blended using a spatula prior to addition of the 4 g mixture into a Thermo Scientific HaakeTM Minilab Micro Compounder (Thermo Fisher Scientific, Waltham, Massachusetts). The Minilab unit is equipped with conical, corotating screws (screw diameter narrowing from 14 to 5 mm, with a length of 11 cm), R
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
Figure 1. Hydrogen bonding synthons (indicated by dashed lines) for (a) the CBZ–NCT cocrystal, and (b) CBZ, polymorphic form III.
and with a control valve for adjusting material residence time. Preliminary tests showed that for this model system, 20% (w/w) polymer load provided a sufficient level of the matrix material, suitable for processing with the extruder, which has a torque limit of 500 N cm. The matrix content of the MAC product plays an important role on processing conditions. The lower the matrix content, the higher the induced shear stress during processing and the higher the exerted torque in the extruder. Therefore, the minimum matrix content is controlled, to a good extent, by the highest acceptable torque for the extrusion unit. In this study, the operating conditions were chosen to enable processing with high mixing intensity (high but acceptable shear stress) of the solid drug and solid coformer, in order to promote cocrystallization. This situation involves the use of a processing temperature where the two cocrystal-forming components exist in the solid state, resulting in intimate mixing conditions similar to those produced by cogrinding. The liquid/softened matrix provides an intervening medium whose effect is analogous to that of a catalyzing solvent for cocrystal production by grinding. Under these conditions, the highest screw speed (shear forces) allowable is set by the torque limit of the extruder. Using these guidelines, MAC batches were processed at a barrel temperature of 115◦ C, with a screw speed of 75 rpm, and residence time of 20 min. It should be pointed out that the residence time in most extruders is controlled by barrel length, screw design, and either screw speed or feed rate, depending on the material feed type. As a result, feed materials experience DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
a distribution of the residence time (the time it takes to pass through the extruder die).14 The control valve in the extruder of this study provides direct control of residence time by the operator. The residence time chosen in the MAC experiments reported here corresponds to the upper end of residence time distributions in larger scale extruders.14 After processing, the obtained extrudate was gently ground with mortar and pestle and screened through a US 100 mesh sieve (aperture size of 152 :m) in preparation for further analysis. The CBZ–NCT cocrystals used as reference for the study were crystallized from solution. The RCC consisted of native cocrystals, free from the effects of any processing. Equimolar amounts of CBZ and NCT powder were suspended in acetonitrile under heat and constant stirring until a clear solution was obtained. The resulting solution was poured onto a watch glass, allowing the solvent to evaporate overnight. Cocrystals obtained from the solution were gently ground with a mortar and pestle and screened through a US 100 mesh sieve (aperture size of 152 :m) in preparation for further analysis. Cocrystallization of CBZ and NCT was confirmed by differential scanning calorimetry (DSC) and powder X-ray diffraction (PXRD), using the reference melting point15 and X-ray diffraction pattern.16 The objective of this study was to evaluate the separate effects of (a) the presence of the matrix material when combined with the cocrystal, and (b) the effect of the extrusion process itself. Therefore, the MAC product was compared with the RCC, as well as to a representative physical mixture (PM) of the RCC and Soluplus (in the same proportions as in the MAC product), during product characterization. The MAC product was analyzed by DSC, Fourier transform infrared spectroscopy (FTIR), and PXRD in order to make qualitative and quantitative assessments of cocrystal quality. The criteria used for cocrystal quality were the efficiency of cocrystallization, that is, the completion of the CBZ–NCT reaction, and the quality of the obtained cocrystals based on the sharpness of the PXRD diffractograms and DSC melting endotherms. Thermal analysis was performed in a PerkinElmer DSC 7 (Waltham, MA) equipped with a refrigerated cooling unit (PerkinElmer Intercooler 1) and analyzed using PyrisTM data analysis software. Temperature and heat flow measurements were calibrated at 10◦ C/min using a NIST-traceable indium standard. Samples (2–4 mg) were run in nonhermetically sealed aluminum pans. The heating rate used was 10◦ C/min, and all experiments were performed under dry nitrogen purge at 20 mL/min. Each measurement was performed in triplicate, and onset of melting temperature (Tm,onset ) and enthalpy of fusion (Hf ) values were taken as the average of three measurements. The FTIR experiments were performed in a Nicolet MagnaIR 750 (Thermo Fisher Scientific) equipped with a HeNe laser (633 nm), deuterated triglycine sulfate detector, and a calcium fluoride beam splitter. Experiments were carried out using OMNIC software version 7.3, and data were collected as beam absorbance from 96 coadded sample scans through a frequency range of 4000–400 cm−1 , with a resolution of 4 cm−1 . Powder X-ray diffraction was performed in a Shimadzu XRD6000 (Kyoto, Japan) X-Ray diffractometer, operating at a voltage of 40 kV and amperage of 30 mA, and using a Cu target to produce X-rays. Each sieved sample was loaded evenly onto an aluminum sample holder and scanned at 2◦ /min, with a step size of 0.02◦ , from 5◦ to 30◦ 22 angle. The [111] crystal face of Si was used as an equipment calibration standard. DOI 10.1002/jps.23983
3
Because of the conversion to the dihydrate form of CBZ when exposed to an aqueous environment, it is important for this type of study to have a rapid analysis method. Specifically, a method capable of quickly determining the concentration of dissolved CBZ in the collected suspension aliquots, such that the extent of supersaturation produced when dissolving the cocrystal can be quantified. Dissolution data were obtained by suspending excess sample into the barrel of a 20 mL syringe filled with 18 mL of deionized water. The suspensions were agitated by affixing the syringes to a rotating shaker set at 20 rpm. At the desired time points, aliquots of approximately 1.5 mL of the suspensions were collected and quickly filtered through a 0.45 :m surfactant-free cellulose acetate membrane. A volume of 1 mL from each filtered aliquot was taken and diluted 1:50 with deionized water for UV absorbance measurements. The diluted solution was subjected to UV absorbance analysis at a wavelength of 310 nm, at which only CBZ absorbs, hence abating potential interference from NCT and Soluplus. Absorbance was converted to CBZ concentration using a calibration curve with R2 value >0.99. This procedure enabled rapid and accurate sampling and concentration measurement of CBZ in solution. This is particularly relevant to the study of the CBZ–NCT cocrystal because of its significantly greater solubility relative to that of CBZ dihydrate. Generating a supersaturated CBZ solution by dissolution of the CBZ–NCT cocrystal also gives information on solution stability and the kinetics of any precipitation to lower solubility forms. Concentration versus time dissolution data were taken as the average of three experiments. Cocrystal Phase Quantification In order to fully evaluate the quality of the MAC product, a multivariate analysis method for assessing the cocrystal content was developed. Partial least squares (PLS) regression analysis was performed on the PXRD data, which showed a relatively linear response between the intensity of diffraction peaks and cocrystal content in calibration standards. The application of PLS and its utilization as a method of sample phase quantification have been thoroughly discussed in the literature.17–21 Calibration standards were made from three basic components: the RCC, a 1:1 molar mixture of CBZ and NCT, and Soluplus. These basic components were used to generate the tertiary mixtures, binary mixtures, as well as the three single phases. Each calibration standard was analyzed by PXRD as described above, and diffraction data were fitted using the SAS software (Cary, NC) version 9.3. The number of factors used in the predictive model was guided by cross-validation of the calibration data, which produces the root mean predicted residual sums of squares (PRESS) for each number of factors used. For the standards used in this study, it was determined that the first seven factors accounted for almost all of the variation in the data without incurring into issues associated with model overfitting. The accuracy of the predictive model generated in SAS was evaluated for all three components by plotting actual versus predicted values of ten validation standards. R
RESULTS AND DISCUSSION Analysis of the MAC product by DSC, FTIR, and PXRD indicate a predominantly cocrystalline system dispersed in the matrix, with the absence of unreacted parent components of the Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
4
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 2. DSC thermograms of (a) MAC product. (b) Reference (RCC– Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) PM of the individual components, CBZ, NCT, and Soluplus. With the exception of thermogram (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal–Soluplus) MAC product.
Figure 4. Comparison of the PXRD profiles among CBZ– NCT/Soluplus formulations. (a) MAC product. (b) Reference (RCC– Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) Physical (ternary) mixture of the individual components, CBZ, NCT, and Soluplus. With the exception of diffractogram (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal:Soluplus) MAC product. Table 1. Comparison of Thermal Data Among CBZ–NCT Formulations
Tm,onset (◦ C) Hf (J/g)
RCC
RCC-Soluplus (80:20) PM
MAC Product
159.3 (0.6) 143.4 (0.4)
155.7 (0.6) 116.9 (0.9)
155.6 (0.3) 110.8 (0.7)
SDs are represented in parentheses.
Figure 3. Comparison of FTIR amide bonding region among (a) MAC product. (b) Reference (RCC–Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) PM of the individual components, CBZ, NCT, and Soluplus. With the exception of spectrum (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal:Soluplus) MAC product.
cocrystal as shown in Figures 2–4. The results show great similarity between the MAC product and the PM containing the same proportions of Soluplus (20%, w/w) with the remainder 80% consisting of the RCC. The DSC results in Table 1 and Figure 2 show that the MAC product and the corresponding PM exhibit very similar melting point, which is slightly depressed (∼4◦ C) in relation to that of the RCC. The enthalpy of melting (Hf ) of the MAC product is close (within 5%) to that of the PM. The similar melting point depression effect observed Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
with the MAC product and the PM indicate a similar mixing effect of CBZ and NCT with the softened Soluplus matrix in both cases. These results also suggest that, even if at trace levels, there is some free CBZ and NCT mixed with Soluplus. In general, a somewhat lower Hf , as in the MAC product, suggests a more intimate mixing of the components. Accordingly, it can be expected that the MAC process results in a somewhat higher incorporation of free CBZ and NCT into the matrix material, relative to the PM. However, the fact that the MAC product and PM show a slight difference in Hf with virtually no difference in melting point suggests an additional possibility involving a particle size effect. The presence of small cocrystal particles homogeneously distributed within the matrix in the MAC product is consistent with a lower Hf but similar melting point as that of the PM of RCC and Soluplus. These results show that the presence of Soluplus itself has the most significant effect in the melting of the CBZ–NCT cocrystals and that the MAC process has a minor effect on the properties of the obtained cocrystals, relative to the RCC. The data obtained from FTIR experiments also give no indication of unreacted cocrystal components present in the MAC product (Fig. 3). The RCC, reference PM, and the MAC product, all display characteristic amide hydrogen bonding (C=O···H–N) peaks at 3390 and 3450 cm−1 . These peaks are distinct from that of the three-component PM made by individually mixing the individual components (CBZ, NCT, and Soluplus) in the same proportions as in the MAC product or in the DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Table 2. Percent Variation Accounted for by PLS Factors for the Optimized PLS Model used to Quantify the Composition in the MAC Product
Table 3. Internal Validation Performed for the Optimization of the PLS Model Used to Quantify the Composition in the MAC Product Split-Sample Validation for the Number of Extracted Factors
Percent Variation Accounted for by Partial Least Squares Factors Number of Extracted Factors Root Mean PRESS Number of Extracted Factors 1 2 3 4 5 6 7 8 9 10
Model Effects Current 46.6705 34.9589 7.6282 1.7258 1.6268 1.3655 1.0641 1.4243 1.3224 0.7320
Current
Total
46.6705 81.6293 89.2576 90.9833 62.6102 63.9757 95.0399 96.4642 97.7866 98.5186
48.1361 49.0077 1.7264 0.9087 0.1693 0.0279 0.0198 0.0036 0.0003 0.0001
48.1361 97.1438 98.8702 99.7789 99.9482 99.9762 99.9960 99.9996 99.9999 100.0000
reference (RCC–Soluplus) PM. The latter shows the characteristic amide hydrogen bonding peak at 3470 cm−1 . The downward peak shift from separate component CBZ/NCT/Soluplus mixture to cocrystal and the introduction of a second amide bonding peak confirm that a change in bonding patterns occurred during MAC, forming a cocrystal between the two components. The polymorphic form III of CBZ assumes a dimer homosynthon (Fig. 1b) between amide groups.22 The formation of the CBZ–NCT cocrystal introduces bonding between CBZ and NCT, as well as an alteration of CBZ bonding motifs (Fig. 1a). CBZ molecules in the cocrystal unit cell bond in a chain motif, rather than the relatively stronger dimer bond, hence the peak shift from 3470 to 3450 cm−1 from the single-component crystals to the cocrystalline form. All CBZ–NCT cocrystal spectra were consistent with existing studies, and no evidence of pure CBZ or NCT was observed.23 The PXRD data shown in Figure 4 supports the findings from the DSC and FTIR studies: the extruded product consists of a CBZ–NCT cocrystal phase and the amorphous Soluplus phase, with no evidence of the unreacted (separate crystalline) forms of CBZ and NCT present in the samples. The PXRD profile matches both the RCC and the reference RCC–Soluplus PM in peak location, and shows no evidence of peaks characteristic of the individual components in the PM. The extent of cocrystallization in the MAC product was quantified using the PLS regression model that was validated for each component to have R2 ≥ 0.99 when fit to a line indicating maximum power. The results of the PLS model are shown in Tables 2 and 3 and in Figure 5. The optimized PLS model estimated the composition of the MAC product to be 78.0% (±0.4) CBZ–NCT cocrystal, with 0.3% (±2.9) PM of unconverted CBZ and NCT PM, and 21.6% (±3.2) Soluplus. Both the qualitative and quantitative findings from DSC, FTIR, and PXRD suggest that the CBZ– NCT cocrystal particles embedded in the Soluplus matrix are nearly equivalent in quality to the mixture of RCC produced from solution cocrystallization and Soluplus. These results indicate that MAC is a viable method for continuous production of high-quality cocrystals embedded in a functional matrix. The in vitro dissolution performance of the MAC product presented in Figure 6 shows the separate effects of (a) the presence of the matrix material when physically mixed with the cocrystal, and (b) the effect of the MAC process itself, DOI 10.1002/jps.23983
T2
Prob > T2
7.5993 7.0571 5.1513 1.8725 2.9874 6.3856 1.9859 0 5.3824 7.2696 6.8351
0.1
resulting in the formation of cocrystal particles embedded in a functional matrix (Soluplus). Figure 6 clearly shows that by only transforming CBZ into the much more highly soluble CBZ– NCT cocrystal results in no dissolution advantage at all. Upon dissolution, the CBZ–NCT cocrystal immediately transforms to the hydrated form of CBZ. It is noteworthy from Figure 6 that dissolution of the plain anhydrous form of CBZ actually results in an observable supersaturation effect, relative to the solubility of CBZ dihydrate. These results show that the solubility/dissolution benefits offered by the CBZ–NCT cocrystal are not realized by the mere formation of the cocrystal. The kinetics of CBZ dihydrate formation is such that it is more of an issue during the dissolution of the CBZ–NCT cocrystal than during the dissolution of the anhydrous form of CBZ. The effect of the physical presence of Soluplus is also observable in Figure 6. The dissolution of the PM of RCC and Soluplus (reference PM) results in faster dissolution and higher concentrations than either the individual RCC or anhydrous CBZ. The presence of Soluplus mixed with the RCC results in a supersaturation effect, relative to that of anhydrous CBZ. Furthermore, by mixing 20% (w/w) Soluplus with the RCC resulted in a prolonged period where the concentration of CBZ remained close to the solubility of the anhydrous form. The effect of the MAC process is also clearly observable in Figure 6. The MAC product and the reference (RCC–Soluplus) PM have the same composition. The difference between the two is the HME coprocessing step. Dissolution of the MAC product results in a maximal concentration that is 5.3 times greater (1.092 mg/mL compared with 0.192 mg/mL) than the reference PM. The area under the dissolution curve of the MAC product is 3.3-fold greater than that of the reference PM (70.3 mg min/mL compared with 21.4 mg min/mL). These results are significant, in light of the work performed by McNamara et al.24 where superior in vitro dissolution of a proprietary glutaric acid cocrystal compared with the pure drug correlated to greater exposure of the drug in beagle dogs upon capsule administration. Jung et al.,25 have also demonstrated this type of in vitro/in vivo correlation for indomethacin-saccharin cocrystals. Cocrystallization of CBZ and NCT offers a theoretical solubility/dissolution rate enhancement. This enhancement, however, is unobtainable without the means of a necessary Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
6
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 6. Comparison of the in vitro dissolution profiles among CBZ– NCT/Soluplus formulations.
dissolution rate and degree of solubilization relative to PM of the RCC and Soluplus. The MAC process simultaneously cocrystallizes and enhances the mixing homogeneity of the resulting cocrystal particles dispersed in the hydrophilic matrix. Upon solidification of the matrix, the small cocrystal particles dispersed in the matrix become essentially “frozen” and isolated from other cocrystal particles. Reducing the agglomeration tendency effectively increases the dissolution rate based on the Noyes–Whitney equation.26 A similar extrusion/matrix effect has been found with drugs embedded into mannitol matrices.27 The matrix plays an important role in cocrystal dissolution rates, and can be utilized to the desired effect: using a hydrophilic and highly soluble matrix can accelerate dissolution, while using a hydrophobic, low solubility matrix can slow dissolution for controlled release, for example. The importance of the proper coformer selection for achieving the desired performance of pharmaceutical cocrystals has been demonstrated.28 It is an important means for fine-tuning the dissolution rate and solubility in the MAC process. The results of this study demonstrate that other factors, including the presence of a matrix material and the extrusion process each have separate and significant effects on the dissolution rate of CBZ–NCT cocrystal, which was severely limited by rapid transformation to the lower solubility CBZ dihydrate form.
CONCLUSIONS
Figure 5. Plots of the predicted content of validation samples (solid symbols). Top: Predicted cocrystal (CC) content. Middle: Predicted CBZ + NCT physical mixture content. Bottom: Predicted Soluplus content.
formulation strategy. Addition of a solubilizing and supersaturation promoter agent, such as Soluplus in this case, results in a sizeable increase in the dissolution rate and degree of solubilization of the drug. However, without any changes in composition, the MAC process results in a significantly increased Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
Matrix-assisted cocrystallization is an effective method for producing high-quality cocrystals, while simultaneously incorporating them into a functional matrix material with formulation implications. The CBZ–NCT/Soluplus MAC product, prepared by the melt extrusion of the three components, showed qualitatively and quantitatively, by DSC, FTIR, and PXRD, to be compositionally equivalent to a reference PM of the RCC and Soluplus in the same proportions. Despite the high tendencies of the CBZ–NCT cocrystal to dissociate upon contact with water to form CBZ dihydrate, embedding the cocrystal in 20% (w/w) Soluplus greatly increased its rate and extent of in vitro dissolution. The proposed method for the simultaneous production and formulation of pharmaceutical cocrystals is solvent free, DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
scalable, and amenable to continuous manufacturing, giving it great potential for utilization in commercial production of cocrystal based pharmaceutical products.
ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of the Department of Education’s Graduate Assistance in Areas of National Need (GAANN) program in pharmaceutical engineering, and Lilly Endowment, Inc. Technical support from the The Dane O. Kildsig Center for Pharmaceutical Processing Research (CPPR) is also acknowledged.
REFERENCES 1. Good DJ, Rodriguez-Hornedo N. 2009. Solubility advantage of pharmaceutical cocrystals. Cryst Growth Des 9:2252–2264. 2. Schultheiss N, Newman A. 2009. Pharmaceutical cocrystals and their physicochemical properties. Cryst Growth Des 9:2950–2967. 3. Remenar JF, Morissette SL, Peterson ML, Moulton B, Macphee JM, Guzman HR, Almarsson O. 2003. Crystal engineering of novel cocrystals of a triazole drug with 1,4-dicarboxylic acids. J Am Chem Soc 125:8456–8457. 4. Friscic T, Jones W. 2009. Recent advances in understanding the mechanism of cocrystal formation via grinding. Cryst Growth Des 9:1621–1637. 5. Zhang GGZ, Henry RF, Borchardt TB, Lou X. 2007. Efficient cocrystal screening using solution-mediated phase transformation. J Pharm Sci 96:990–995. 6. Chiarella RA, Davey RJ, Peterson ML. 2007. Making co-crystals— The utility of ternary phase diagrams. Cryst Growth Des 7:1223– 1226. 7. Chadwick K, Davey RJ, Dent G, Pritchard RG, Hunter CA, Musumeci D. 2009. Cocrystallization: A solution chemistry perspective and the case of benzophenone and diphenylamine. Cryst Growth Des 9:1990–1999. 8. Shan N, Toda F, Jones W. 2002. Mechanochemistry and co-crystal formation: Effect of solvent on reaction kinetics. Chem Commun 2372– 2373. 9. Friscic T, Jones W. 2007. Cocrystal architecture and properties: Design and building of chiral and racemic structures by solid–solid reactions. Faraday Disscuss 136:167–178. 10. Nehm SJ, Rodriguez-Spong B, Rodriguez-Hornedo N. 2006. Phase solubility diagrams of cocrystals are explained by solubility product and solution complexation. Cryst Growth Des. 6:592–600. ˜ 11. Medina C, Daurio D, Nagapudi K, Alvarez-Nu´ nez F. 2009. Manufacture of pharmaceutical co-crystals using twin screw extrusion: A solvent-less and scalable process. J Pharm Sci 99:1693–1696.
DOI 10.1002/jps.23983
7
12. BASF. 2010. Technical information—Soluplus. BASF SE, Care Chemicals Division. Pharma Ingredients and Services. Limburgerhof, Germany (www.pharma-ingredients.basf.com) 1–8. 13. Ng YC, Yang Z, McAuley WJ, Qi S. 2013. Stabilisation of amorphous drugs under high humidity using pharmaceutical thin films. Eur J Pharm Biopharm 84:555–565. 14. Steiner R. 2003. Extruder design. In Pharmaceutical extrusion technology; Ghebre-Sellassie I, Martin C, Eds. New York: Marcel Dekker, Inc., pp. 17–36. ˜ F, Rodr´ıguez-Hornedo N. 2007. 15. Seefeldt K, Miller J, Alvarez-Nu´ nez Crystallization pathways and kinetics of carbamazepine–nicotinamide cocrystals from the amorphous state by in situ thermomicroscopy, spectroscopy and calorimetry studies. J Pharm Sci 96:1147–1158. 16. Fleischman SG, Kuduva SS, McMahon JA, Moulton B, Bailey Walsh RD, Rodr´ıguez-Hornedo N, Zaworotko MJ. 2003. Crystal engineering of the composition of pharmaceutical phases: Multiple-component crystalline solids involving carbamazepine. Cryst Growth Des 3:909–919. 17. Wold S, Ruhe A, Wold H, Dunn WJ. 1984. The collinearity problem in linear regression. The partial least squares (pls) approach to generalized inverses. SIAM J Sci. Stat. Comput. 5:735–743. 18. Geladi P, Kowalski BR. 1986. Partial least-squares regression: A tutorial. Anal Chim Acta 185:1–17. 19. Wold S, Sjostrom M, Eriksson L. 2001. Pls-regression: A basic tool of chemometrics. Chemom Intell Lab Sys 58:109–130. 20. Salari A, Young RE. 1998. Application of attenuated total reflectance ftir spectroscopy to the analysis of mixtures of pharmaceutical polymorphs. Int J Pharm 163:157–166. 21. Tobias RD. 1995. An introduction to partial least squares regression. Analysis 24:1250–1257. 22. Himes VL, Mighell AD, De Camp WH. 1981. Structure of carbamazepine: 5h-dibenz b.F azepine-5-carboxamide. Acta Cryst 37:2242– 2245. 23. Porter WW, Elie SC, Matzger AJ. 2008. Polymorphism in carbamazepine cocrystals. Cryst Growth Des 8:14–16. 24. McNamara DP, Childs SL, Giordano J, Iarriccio A, Cassidy J, Shet MS, Mannion R, O’Donnell E, Park A. 2006. Use of a glutaric acid cocrystal to improve oral bioavailability of a low solubility API. Pharm Res 23:1888–1897. 25. Jung M-S, Kim J-S, Kim M-S, Alhalaweh A, Cho W, Hwang S-J, Velaga SP. 2010. Bioavailability of indomethacin-saccharin cocrystals. J Pharm Pharmacol 62:1560–1568. 26. Noyes AA, Whitney WR. 1897. The rate of solution of solid substances in their own solutions. J Am Chem Soc 19:930–934. 27. Thommes M, Ely DR, Carvajal MT, Pinal R. 2011. Improvement of the dissolution rate of poorly soluble drugs by solid crystal suspensions. Mol Pharm 8:727–735. 28. Cheney ML, Weyna DR, Shan N, Hanna M, Wojtas L, Zaworotko MJ. 2011. Coformer selection in pharmaceutical cocrystal development: A case study of a meloxicam aspirin cocrystal that exhibits enhanced solubility and pharmacokinetics. J Pharm Sci 100:2172–2181.
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
Matrix-Assisted Cocrystallization: The Simultaneous Production and Formulation of Pharmaceutical Cocrystals by Hot-Melt Extrusion KEVIN BOKSA, ANDREW OTTE, RODOLFO PINAL Department of Industrial and Physical Pharmacy, Purdue University, West Lafayette, Indiana Received 1 February 2014; accepted 25 March 2014 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23983 ABSTRACT: A novel method for the simultaneous production and formulation of pharmaceutical cocrystals, matrix-assisted cocrystallization (MAC), is presented. Hot-melt extrusion (HME) is used to create cocrystals by coprocessing the drug and coformer in the presence of a matrix material. Carbamazepine (CBZ), nicotinamide (NCT), and Soluplus were used as a model drug, coformer, and matrix, respectively. The MAC product containing 80:20 (w/w) cocrystal:matrix was characterized by differential scanning calorimetry, Fourier transform infrared spectroscopy, and powder X-ray diffraction. A partial least squares (PLS) regression model was developed for quantifying the efficiency of cocrystal formation. The MAC product was estimated to be 78% (w/w) cocrystal (theoretical 80%), with approximately 0.3% mixture of free (unreacted) CBZ and NCT, and 21.6% Soluplus (theoretical 20%) with the PLS model. A physical mixture (PM) of a reference cocrystal (RCC), prepared by precipitation from solution, and Soluplus resulted in faster dissolution relative to the pure RCC. However, the MAC product with the exact same composition resulted in considerably faster dissolution and higher maximum concentration (∼five-fold) than those of the PM. The MAC product consists of high-quality cocrystals embedded in a matrix. The processing aspect of MAC plays a major role on the faster dissolution observed. The MAC approach offers a scalable process, suitable for the continuous manufacturing and C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci formulation of pharmaceutical cocrystals. Keywords: cocrystals; matrix; formulation; extrusion; continuous manufacturing; dissolution; supersaturation; crystal engineering; polymers; crystallization; solid dispersions R
INTRODUCTION Among the different formulation strategies for addressing the low systemic exposure of poorly soluble compounds administered by the oral route, cocrystallization is a promising approach that remains relatively underexplored in terms of formulation improvement. Pharmaceutical cocrystals, defined as solid-state stoichiometric molecular complexes of a drug molecule and a complementary coformer molecule, have demonstrated the ability to improve the solubility and physical stability of drugs.1–3 Nevertheless, their applicability in the pharmaceutical industry remains limited due in good measure to the lack of a suitable large-scale production method. Established cocrystal production methods rely on the generation of an intermediate phase with increased molecular mobility, such as the liquid or amorphous state, from which the components cocrystallize.4–6 Such methods include cocrystallization from a eutectic melt of the cocrystal components, cocrystallization via neat or liquid-assisted grinding (where the liquid acts as a catalyst), and cocrystallization by precipitation from solution. While each of these methods is capable of generating cocrystals, each also poses some limitations. Thermal methods involving melting require elevated temperatures that can compromise the integrity of thermolabile compounds. Besides being energetically inefficient, mechanical energy input methods like grinding can induce amorphization and their effectiveness tends to be limited without the use of a catalytic solvent. Finally, in methods
Correspondence to: Rodolfo Pinal (Telephone: +765-496-6247; Fax: +765-4946545; E-mail: [email protected]) Journal of Pharmaceutical Sciences
C 2014 Wiley Periodicals, Inc. and the American Pharmacists Association
based on precipitation from solution, the location of the system in the phase diagram is critical to the cocrystallization process and to the attributes of the cocrystals produced,7 requiring both dynamic and precise control of supersaturation levels of the concentration of the component concentrations as cocrystallization takes place. This type of coordinated control can be difficult to maintain.6,8–10 A scalable production method for pharmaceutical cocrystals that draws on the advantages of existing methods, such as the control of stoichiometry offered by the thermal and energy input methods, along with the high crystal quality offered by the solution based methods, would be of significant valuable to pharmaceutical scientists during preformulation and formulation development. An equally important aspect for realizing the full potential of pharmaceutical cocrystals is the availability of a robust, scalable production method that effectively addresses the drawbacks (such as the potential chemical and physical stability issues mentioned above) associated with the cocrystal formation methods. In response to the need for an improved process, matrixassisted cocrystallization (MAC) is introduced here as a novel method of cocrystal production. Medina et al.11 have shown that hot-melt extrusion (HME) can be used as an effective cogrinding process to create cocrystals. The MAC approach utilizes HME to produce cocrystals embedded in a formulation matrix. Briefly, equimolar quantities of drug and coformer are mixed with a matrix material in the solid state prior to feeding the mixture into a hot-melt extruder. The extruder is set at a temperature where only the matrix material is made fluid, either by softening or by actual melting. Cocrystallization occurs during the extrusion process, induced by the intimate mixing and grinding of the components in the softened (or liquefied) matrix. The cocrystal particles formed this way become embedded in
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
1
2
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
the matrix material, which solidifies upon exiting the extruder. The end product of the MAC process, termed MAC product in this report, is a matrix-embedded cocrystal with the level of controlled stoichiometry offered by both thermal and mechanical cocrystal production methods. The MAC product is a formulated cocrystal where the matrix material plays a double role. During the HME process, the matrix plays role analogous to that of a catalytic solvent. In the final extrudate, the matrix is a functional component of the formulation. In MAC, the use of the melted/softened matrix phase promotes intimate mixing, while reducing excessive shear stresses produced when subjecting dry solid materials to extrusion. The reduced shear stress of MAC abates potential crystal damage. Producing cocrystal particles in functional matrices makes the physicochemical properties of the matrix as important as those of the cocrystal itself. Therefore, selection of the matrix material can be exploited to impart additional functionality into the formulated cocrystal, such as improved flowability, compaction, drug release kinetics, and so on. Thus, MAC serves as a method of simultaneous production and formulation of pharmaceutical cocrystals. In this report, we examine the ability of the MAC process to produce high-quality cocrystals. We also present the results of characterization studies comparing the physical attributes of the MAC product with those of a reference cocrystal (RCC) material, that is, the same cocrystal obtained by the careful implementation of the solution cocrystallization method. The study also includes a comparison of the in vitro dissolution performance between the MAC product and the RCC.
MATERIALS AND METHODS Carbamazepine (CBZ) and nicotinamide (NCT) were used as the drug and coformer, respectively, for the model cocrystal system of this study. Figure 1 shows the structures of the CBZ–NCT cocrystal and that of crystalline CBZ. CBZ is a BCS class II compound, with dissolution rate-limited absorption. One important aspect is that the anhydrous form of crystalline CBZ rapidly transforms into the dihydrate form during dissolution in aqueous media. This poses a problem because the transformation to the dihydrate results in a physical form that is less soluble, with considerably slower dissolution than the original anhydrous form. NCT was used as coformer for this study because it has been reported to form a cocrystal with CBZ whose solubility is 152-fold greater than that of the CBZ dihydrate.1 Soluplus was utilized as the matrix agent for cocrystal formation because of its relatively low glass transition temperature (∼72◦ C), allowing for a convenient processing window during extrusion. In addition, Soluplus has been shown to solubilize drugs and prevent their precipitation from supersaturated solutions.12,13 CBZ (CAS #298–46–4, Fig. 1b) was purchased from Alfa Aesar (Ward Hill, MA); NCT (CAS #98–92–0) was purchased from Sigma–Aldrich (St. Louis, MO); and Soluplus (CAS #402932–23–4) was obtained from BASF (Florham Park, NJ). All materials were used as received. In a typical MAC experiment, 3.2 g of a 1:1 molar ratio (66:34 mass ratio) CBZ–NCT mixture and 0.8 g Soluplus were blended using a spatula prior to addition of the 4 g mixture into a Thermo Scientific HaakeTM Minilab Micro Compounder (Thermo Fisher Scientific, Waltham, Massachusetts). The Minilab unit is equipped with conical, corotating screws (screw diameter narrowing from 14 to 5 mm, with a length of 11 cm), R
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
Figure 1. Hydrogen bonding synthons (indicated by dashed lines) for (a) the CBZ–NCT cocrystal, and (b) CBZ, polymorphic form III.
and with a control valve for adjusting material residence time. Preliminary tests showed that for this model system, 20% (w/w) polymer load provided a sufficient level of the matrix material, suitable for processing with the extruder, which has a torque limit of 500 N cm. The matrix content of the MAC product plays an important role on processing conditions. The lower the matrix content, the higher the induced shear stress during processing and the higher the exerted torque in the extruder. Therefore, the minimum matrix content is controlled, to a good extent, by the highest acceptable torque for the extrusion unit. In this study, the operating conditions were chosen to enable processing with high mixing intensity (high but acceptable shear stress) of the solid drug and solid coformer, in order to promote cocrystallization. This situation involves the use of a processing temperature where the two cocrystal-forming components exist in the solid state, resulting in intimate mixing conditions similar to those produced by cogrinding. The liquid/softened matrix provides an intervening medium whose effect is analogous to that of a catalyzing solvent for cocrystal production by grinding. Under these conditions, the highest screw speed (shear forces) allowable is set by the torque limit of the extruder. Using these guidelines, MAC batches were processed at a barrel temperature of 115◦ C, with a screw speed of 75 rpm, and residence time of 20 min. It should be pointed out that the residence time in most extruders is controlled by barrel length, screw design, and either screw speed or feed rate, depending on the material feed type. As a result, feed materials experience DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
a distribution of the residence time (the time it takes to pass through the extruder die).14 The control valve in the extruder of this study provides direct control of residence time by the operator. The residence time chosen in the MAC experiments reported here corresponds to the upper end of residence time distributions in larger scale extruders.14 After processing, the obtained extrudate was gently ground with mortar and pestle and screened through a US 100 mesh sieve (aperture size of 152 :m) in preparation for further analysis. The CBZ–NCT cocrystals used as reference for the study were crystallized from solution. The RCC consisted of native cocrystals, free from the effects of any processing. Equimolar amounts of CBZ and NCT powder were suspended in acetonitrile under heat and constant stirring until a clear solution was obtained. The resulting solution was poured onto a watch glass, allowing the solvent to evaporate overnight. Cocrystals obtained from the solution were gently ground with a mortar and pestle and screened through a US 100 mesh sieve (aperture size of 152 :m) in preparation for further analysis. Cocrystallization of CBZ and NCT was confirmed by differential scanning calorimetry (DSC) and powder X-ray diffraction (PXRD), using the reference melting point15 and X-ray diffraction pattern.16 The objective of this study was to evaluate the separate effects of (a) the presence of the matrix material when combined with the cocrystal, and (b) the effect of the extrusion process itself. Therefore, the MAC product was compared with the RCC, as well as to a representative physical mixture (PM) of the RCC and Soluplus (in the same proportions as in the MAC product), during product characterization. The MAC product was analyzed by DSC, Fourier transform infrared spectroscopy (FTIR), and PXRD in order to make qualitative and quantitative assessments of cocrystal quality. The criteria used for cocrystal quality were the efficiency of cocrystallization, that is, the completion of the CBZ–NCT reaction, and the quality of the obtained cocrystals based on the sharpness of the PXRD diffractograms and DSC melting endotherms. Thermal analysis was performed in a PerkinElmer DSC 7 (Waltham, MA) equipped with a refrigerated cooling unit (PerkinElmer Intercooler 1) and analyzed using PyrisTM data analysis software. Temperature and heat flow measurements were calibrated at 10◦ C/min using a NIST-traceable indium standard. Samples (2–4 mg) were run in nonhermetically sealed aluminum pans. The heating rate used was 10◦ C/min, and all experiments were performed under dry nitrogen purge at 20 mL/min. Each measurement was performed in triplicate, and onset of melting temperature (Tm,onset ) and enthalpy of fusion (Hf ) values were taken as the average of three measurements. The FTIR experiments were performed in a Nicolet MagnaIR 750 (Thermo Fisher Scientific) equipped with a HeNe laser (633 nm), deuterated triglycine sulfate detector, and a calcium fluoride beam splitter. Experiments were carried out using OMNIC software version 7.3, and data were collected as beam absorbance from 96 coadded sample scans through a frequency range of 4000–400 cm−1 , with a resolution of 4 cm−1 . Powder X-ray diffraction was performed in a Shimadzu XRD6000 (Kyoto, Japan) X-Ray diffractometer, operating at a voltage of 40 kV and amperage of 30 mA, and using a Cu target to produce X-rays. Each sieved sample was loaded evenly onto an aluminum sample holder and scanned at 2◦ /min, with a step size of 0.02◦ , from 5◦ to 30◦ 22 angle. The [111] crystal face of Si was used as an equipment calibration standard. DOI 10.1002/jps.23983
3
Because of the conversion to the dihydrate form of CBZ when exposed to an aqueous environment, it is important for this type of study to have a rapid analysis method. Specifically, a method capable of quickly determining the concentration of dissolved CBZ in the collected suspension aliquots, such that the extent of supersaturation produced when dissolving the cocrystal can be quantified. Dissolution data were obtained by suspending excess sample into the barrel of a 20 mL syringe filled with 18 mL of deionized water. The suspensions were agitated by affixing the syringes to a rotating shaker set at 20 rpm. At the desired time points, aliquots of approximately 1.5 mL of the suspensions were collected and quickly filtered through a 0.45 :m surfactant-free cellulose acetate membrane. A volume of 1 mL from each filtered aliquot was taken and diluted 1:50 with deionized water for UV absorbance measurements. The diluted solution was subjected to UV absorbance analysis at a wavelength of 310 nm, at which only CBZ absorbs, hence abating potential interference from NCT and Soluplus. Absorbance was converted to CBZ concentration using a calibration curve with R2 value >0.99. This procedure enabled rapid and accurate sampling and concentration measurement of CBZ in solution. This is particularly relevant to the study of the CBZ–NCT cocrystal because of its significantly greater solubility relative to that of CBZ dihydrate. Generating a supersaturated CBZ solution by dissolution of the CBZ–NCT cocrystal also gives information on solution stability and the kinetics of any precipitation to lower solubility forms. Concentration versus time dissolution data were taken as the average of three experiments. Cocrystal Phase Quantification In order to fully evaluate the quality of the MAC product, a multivariate analysis method for assessing the cocrystal content was developed. Partial least squares (PLS) regression analysis was performed on the PXRD data, which showed a relatively linear response between the intensity of diffraction peaks and cocrystal content in calibration standards. The application of PLS and its utilization as a method of sample phase quantification have been thoroughly discussed in the literature.17–21 Calibration standards were made from three basic components: the RCC, a 1:1 molar mixture of CBZ and NCT, and Soluplus. These basic components were used to generate the tertiary mixtures, binary mixtures, as well as the three single phases. Each calibration standard was analyzed by PXRD as described above, and diffraction data were fitted using the SAS software (Cary, NC) version 9.3. The number of factors used in the predictive model was guided by cross-validation of the calibration data, which produces the root mean predicted residual sums of squares (PRESS) for each number of factors used. For the standards used in this study, it was determined that the first seven factors accounted for almost all of the variation in the data without incurring into issues associated with model overfitting. The accuracy of the predictive model generated in SAS was evaluated for all three components by plotting actual versus predicted values of ten validation standards. R
RESULTS AND DISCUSSION Analysis of the MAC product by DSC, FTIR, and PXRD indicate a predominantly cocrystalline system dispersed in the matrix, with the absence of unreacted parent components of the Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
4
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 2. DSC thermograms of (a) MAC product. (b) Reference (RCC– Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) PM of the individual components, CBZ, NCT, and Soluplus. With the exception of thermogram (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal–Soluplus) MAC product.
Figure 4. Comparison of the PXRD profiles among CBZ– NCT/Soluplus formulations. (a) MAC product. (b) Reference (RCC– Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) Physical (ternary) mixture of the individual components, CBZ, NCT, and Soluplus. With the exception of diffractogram (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal:Soluplus) MAC product. Table 1. Comparison of Thermal Data Among CBZ–NCT Formulations
Tm,onset (◦ C) Hf (J/g)
RCC
RCC-Soluplus (80:20) PM
MAC Product
159.3 (0.6) 143.4 (0.4)
155.7 (0.6) 116.9 (0.9)
155.6 (0.3) 110.8 (0.7)
SDs are represented in parentheses.
Figure 3. Comparison of FTIR amide bonding region among (a) MAC product. (b) Reference (RCC–Soluplus) PM. (c) Reference CBZ–NCT cocrystal (RCC). (d) PM of the individual components, CBZ, NCT, and Soluplus. With the exception of spectrum (c) (100% cocrystal), all samples have the same proportions as the 80:20 (cocrystal:Soluplus) MAC product.
cocrystal as shown in Figures 2–4. The results show great similarity between the MAC product and the PM containing the same proportions of Soluplus (20%, w/w) with the remainder 80% consisting of the RCC. The DSC results in Table 1 and Figure 2 show that the MAC product and the corresponding PM exhibit very similar melting point, which is slightly depressed (∼4◦ C) in relation to that of the RCC. The enthalpy of melting (Hf ) of the MAC product is close (within 5%) to that of the PM. The similar melting point depression effect observed Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
with the MAC product and the PM indicate a similar mixing effect of CBZ and NCT with the softened Soluplus matrix in both cases. These results also suggest that, even if at trace levels, there is some free CBZ and NCT mixed with Soluplus. In general, a somewhat lower Hf , as in the MAC product, suggests a more intimate mixing of the components. Accordingly, it can be expected that the MAC process results in a somewhat higher incorporation of free CBZ and NCT into the matrix material, relative to the PM. However, the fact that the MAC product and PM show a slight difference in Hf with virtually no difference in melting point suggests an additional possibility involving a particle size effect. The presence of small cocrystal particles homogeneously distributed within the matrix in the MAC product is consistent with a lower Hf but similar melting point as that of the PM of RCC and Soluplus. These results show that the presence of Soluplus itself has the most significant effect in the melting of the CBZ–NCT cocrystals and that the MAC process has a minor effect on the properties of the obtained cocrystals, relative to the RCC. The data obtained from FTIR experiments also give no indication of unreacted cocrystal components present in the MAC product (Fig. 3). The RCC, reference PM, and the MAC product, all display characteristic amide hydrogen bonding (C=O···H–N) peaks at 3390 and 3450 cm−1 . These peaks are distinct from that of the three-component PM made by individually mixing the individual components (CBZ, NCT, and Soluplus) in the same proportions as in the MAC product or in the DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Table 2. Percent Variation Accounted for by PLS Factors for the Optimized PLS Model used to Quantify the Composition in the MAC Product
Table 3. Internal Validation Performed for the Optimization of the PLS Model Used to Quantify the Composition in the MAC Product Split-Sample Validation for the Number of Extracted Factors
Percent Variation Accounted for by Partial Least Squares Factors Number of Extracted Factors Root Mean PRESS Number of Extracted Factors 1 2 3 4 5 6 7 8 9 10
Model Effects Current 46.6705 34.9589 7.6282 1.7258 1.6268 1.3655 1.0641 1.4243 1.3224 0.7320
Current
Total
46.6705 81.6293 89.2576 90.9833 62.6102 63.9757 95.0399 96.4642 97.7866 98.5186
48.1361 49.0077 1.7264 0.9087 0.1693 0.0279 0.0198 0.0036 0.0003 0.0001
48.1361 97.1438 98.8702 99.7789 99.9482 99.9762 99.9960 99.9996 99.9999 100.0000
reference (RCC–Soluplus) PM. The latter shows the characteristic amide hydrogen bonding peak at 3470 cm−1 . The downward peak shift from separate component CBZ/NCT/Soluplus mixture to cocrystal and the introduction of a second amide bonding peak confirm that a change in bonding patterns occurred during MAC, forming a cocrystal between the two components. The polymorphic form III of CBZ assumes a dimer homosynthon (Fig. 1b) between amide groups.22 The formation of the CBZ–NCT cocrystal introduces bonding between CBZ and NCT, as well as an alteration of CBZ bonding motifs (Fig. 1a). CBZ molecules in the cocrystal unit cell bond in a chain motif, rather than the relatively stronger dimer bond, hence the peak shift from 3470 to 3450 cm−1 from the single-component crystals to the cocrystalline form. All CBZ–NCT cocrystal spectra were consistent with existing studies, and no evidence of pure CBZ or NCT was observed.23 The PXRD data shown in Figure 4 supports the findings from the DSC and FTIR studies: the extruded product consists of a CBZ–NCT cocrystal phase and the amorphous Soluplus phase, with no evidence of the unreacted (separate crystalline) forms of CBZ and NCT present in the samples. The PXRD profile matches both the RCC and the reference RCC–Soluplus PM in peak location, and shows no evidence of peaks characteristic of the individual components in the PM. The extent of cocrystallization in the MAC product was quantified using the PLS regression model that was validated for each component to have R2 ≥ 0.99 when fit to a line indicating maximum power. The results of the PLS model are shown in Tables 2 and 3 and in Figure 5. The optimized PLS model estimated the composition of the MAC product to be 78.0% (±0.4) CBZ–NCT cocrystal, with 0.3% (±2.9) PM of unconverted CBZ and NCT PM, and 21.6% (±3.2) Soluplus. Both the qualitative and quantitative findings from DSC, FTIR, and PXRD suggest that the CBZ– NCT cocrystal particles embedded in the Soluplus matrix are nearly equivalent in quality to the mixture of RCC produced from solution cocrystallization and Soluplus. These results indicate that MAC is a viable method for continuous production of high-quality cocrystals embedded in a functional matrix. The in vitro dissolution performance of the MAC product presented in Figure 6 shows the separate effects of (a) the presence of the matrix material when physically mixed with the cocrystal, and (b) the effect of the MAC process itself, DOI 10.1002/jps.23983
T2
Prob > T2
7.5993 7.0571 5.1513 1.8725 2.9874 6.3856 1.9859 0 5.3824 7.2696 6.8351
0.1
resulting in the formation of cocrystal particles embedded in a functional matrix (Soluplus). Figure 6 clearly shows that by only transforming CBZ into the much more highly soluble CBZ– NCT cocrystal results in no dissolution advantage at all. Upon dissolution, the CBZ–NCT cocrystal immediately transforms to the hydrated form of CBZ. It is noteworthy from Figure 6 that dissolution of the plain anhydrous form of CBZ actually results in an observable supersaturation effect, relative to the solubility of CBZ dihydrate. These results show that the solubility/dissolution benefits offered by the CBZ–NCT cocrystal are not realized by the mere formation of the cocrystal. The kinetics of CBZ dihydrate formation is such that it is more of an issue during the dissolution of the CBZ–NCT cocrystal than during the dissolution of the anhydrous form of CBZ. The effect of the physical presence of Soluplus is also observable in Figure 6. The dissolution of the PM of RCC and Soluplus (reference PM) results in faster dissolution and higher concentrations than either the individual RCC or anhydrous CBZ. The presence of Soluplus mixed with the RCC results in a supersaturation effect, relative to that of anhydrous CBZ. Furthermore, by mixing 20% (w/w) Soluplus with the RCC resulted in a prolonged period where the concentration of CBZ remained close to the solubility of the anhydrous form. The effect of the MAC process is also clearly observable in Figure 6. The MAC product and the reference (RCC–Soluplus) PM have the same composition. The difference between the two is the HME coprocessing step. Dissolution of the MAC product results in a maximal concentration that is 5.3 times greater (1.092 mg/mL compared with 0.192 mg/mL) than the reference PM. The area under the dissolution curve of the MAC product is 3.3-fold greater than that of the reference PM (70.3 mg min/mL compared with 21.4 mg min/mL). These results are significant, in light of the work performed by McNamara et al.24 where superior in vitro dissolution of a proprietary glutaric acid cocrystal compared with the pure drug correlated to greater exposure of the drug in beagle dogs upon capsule administration. Jung et al.,25 have also demonstrated this type of in vitro/in vivo correlation for indomethacin-saccharin cocrystals. Cocrystallization of CBZ and NCT offers a theoretical solubility/dissolution rate enhancement. This enhancement, however, is unobtainable without the means of a necessary Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
6
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
Figure 6. Comparison of the in vitro dissolution profiles among CBZ– NCT/Soluplus formulations.
dissolution rate and degree of solubilization relative to PM of the RCC and Soluplus. The MAC process simultaneously cocrystallizes and enhances the mixing homogeneity of the resulting cocrystal particles dispersed in the hydrophilic matrix. Upon solidification of the matrix, the small cocrystal particles dispersed in the matrix become essentially “frozen” and isolated from other cocrystal particles. Reducing the agglomeration tendency effectively increases the dissolution rate based on the Noyes–Whitney equation.26 A similar extrusion/matrix effect has been found with drugs embedded into mannitol matrices.27 The matrix plays an important role in cocrystal dissolution rates, and can be utilized to the desired effect: using a hydrophilic and highly soluble matrix can accelerate dissolution, while using a hydrophobic, low solubility matrix can slow dissolution for controlled release, for example. The importance of the proper coformer selection for achieving the desired performance of pharmaceutical cocrystals has been demonstrated.28 It is an important means for fine-tuning the dissolution rate and solubility in the MAC process. The results of this study demonstrate that other factors, including the presence of a matrix material and the extrusion process each have separate and significant effects on the dissolution rate of CBZ–NCT cocrystal, which was severely limited by rapid transformation to the lower solubility CBZ dihydrate form.
CONCLUSIONS
Figure 5. Plots of the predicted content of validation samples (solid symbols). Top: Predicted cocrystal (CC) content. Middle: Predicted CBZ + NCT physical mixture content. Bottom: Predicted Soluplus content.
formulation strategy. Addition of a solubilizing and supersaturation promoter agent, such as Soluplus in this case, results in a sizeable increase in the dissolution rate and degree of solubilization of the drug. However, without any changes in composition, the MAC process results in a significantly increased Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
Matrix-assisted cocrystallization is an effective method for producing high-quality cocrystals, while simultaneously incorporating them into a functional matrix material with formulation implications. The CBZ–NCT/Soluplus MAC product, prepared by the melt extrusion of the three components, showed qualitatively and quantitatively, by DSC, FTIR, and PXRD, to be compositionally equivalent to a reference PM of the RCC and Soluplus in the same proportions. Despite the high tendencies of the CBZ–NCT cocrystal to dissociate upon contact with water to form CBZ dihydrate, embedding the cocrystal in 20% (w/w) Soluplus greatly increased its rate and extent of in vitro dissolution. The proposed method for the simultaneous production and formulation of pharmaceutical cocrystals is solvent free, DOI 10.1002/jps.23983
RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology
scalable, and amenable to continuous manufacturing, giving it great potential for utilization in commercial production of cocrystal based pharmaceutical products.
ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of the Department of Education’s Graduate Assistance in Areas of National Need (GAANN) program in pharmaceutical engineering, and Lilly Endowment, Inc. Technical support from the The Dane O. Kildsig Center for Pharmaceutical Processing Research (CPPR) is also acknowledged.
REFERENCES 1. Good DJ, Rodriguez-Hornedo N. 2009. Solubility advantage of pharmaceutical cocrystals. Cryst Growth Des 9:2252–2264. 2. Schultheiss N, Newman A. 2009. Pharmaceutical cocrystals and their physicochemical properties. Cryst Growth Des 9:2950–2967. 3. Remenar JF, Morissette SL, Peterson ML, Moulton B, Macphee JM, Guzman HR, Almarsson O. 2003. Crystal engineering of novel cocrystals of a triazole drug with 1,4-dicarboxylic acids. J Am Chem Soc 125:8456–8457. 4. Friscic T, Jones W. 2009. Recent advances in understanding the mechanism of cocrystal formation via grinding. Cryst Growth Des 9:1621–1637. 5. Zhang GGZ, Henry RF, Borchardt TB, Lou X. 2007. Efficient cocrystal screening using solution-mediated phase transformation. J Pharm Sci 96:990–995. 6. Chiarella RA, Davey RJ, Peterson ML. 2007. Making co-crystals— The utility of ternary phase diagrams. Cryst Growth Des 7:1223– 1226. 7. Chadwick K, Davey RJ, Dent G, Pritchard RG, Hunter CA, Musumeci D. 2009. Cocrystallization: A solution chemistry perspective and the case of benzophenone and diphenylamine. Cryst Growth Des 9:1990–1999. 8. Shan N, Toda F, Jones W. 2002. Mechanochemistry and co-crystal formation: Effect of solvent on reaction kinetics. Chem Commun 2372– 2373. 9. Friscic T, Jones W. 2007. Cocrystal architecture and properties: Design and building of chiral and racemic structures by solid–solid reactions. Faraday Disscuss 136:167–178. 10. Nehm SJ, Rodriguez-Spong B, Rodriguez-Hornedo N. 2006. Phase solubility diagrams of cocrystals are explained by solubility product and solution complexation. Cryst Growth Des. 6:592–600. ˜ 11. Medina C, Daurio D, Nagapudi K, Alvarez-Nu´ nez F. 2009. Manufacture of pharmaceutical co-crystals using twin screw extrusion: A solvent-less and scalable process. J Pharm Sci 99:1693–1696.
DOI 10.1002/jps.23983
7
12. BASF. 2010. Technical information—Soluplus. BASF SE, Care Chemicals Division. Pharma Ingredients and Services. Limburgerhof, Germany (www.pharma-ingredients.basf.com) 1–8. 13. Ng YC, Yang Z, McAuley WJ, Qi S. 2013. Stabilisation of amorphous drugs under high humidity using pharmaceutical thin films. Eur J Pharm Biopharm 84:555–565. 14. Steiner R. 2003. Extruder design. In Pharmaceutical extrusion technology; Ghebre-Sellassie I, Martin C, Eds. New York: Marcel Dekker, Inc., pp. 17–36. ˜ F, Rodr´ıguez-Hornedo N. 2007. 15. Seefeldt K, Miller J, Alvarez-Nu´ nez Crystallization pathways and kinetics of carbamazepine–nicotinamide cocrystals from the amorphous state by in situ thermomicroscopy, spectroscopy and calorimetry studies. J Pharm Sci 96:1147–1158. 16. Fleischman SG, Kuduva SS, McMahon JA, Moulton B, Bailey Walsh RD, Rodr´ıguez-Hornedo N, Zaworotko MJ. 2003. Crystal engineering of the composition of pharmaceutical phases: Multiple-component crystalline solids involving carbamazepine. Cryst Growth Des 3:909–919. 17. Wold S, Ruhe A, Wold H, Dunn WJ. 1984. The collinearity problem in linear regression. The partial least squares (pls) approach to generalized inverses. SIAM J Sci. Stat. Comput. 5:735–743. 18. Geladi P, Kowalski BR. 1986. Partial least-squares regression: A tutorial. Anal Chim Acta 185:1–17. 19. Wold S, Sjostrom M, Eriksson L. 2001. Pls-regression: A basic tool of chemometrics. Chemom Intell Lab Sys 58:109–130. 20. Salari A, Young RE. 1998. Application of attenuated total reflectance ftir spectroscopy to the analysis of mixtures of pharmaceutical polymorphs. Int J Pharm 163:157–166. 21. Tobias RD. 1995. An introduction to partial least squares regression. Analysis 24:1250–1257. 22. Himes VL, Mighell AD, De Camp WH. 1981. Structure of carbamazepine: 5h-dibenz b.F azepine-5-carboxamide. Acta Cryst 37:2242– 2245. 23. Porter WW, Elie SC, Matzger AJ. 2008. Polymorphism in carbamazepine cocrystals. Cryst Growth Des 8:14–16. 24. McNamara DP, Childs SL, Giordano J, Iarriccio A, Cassidy J, Shet MS, Mannion R, O’Donnell E, Park A. 2006. Use of a glutaric acid cocrystal to improve oral bioavailability of a low solubility API. Pharm Res 23:1888–1897. 25. Jung M-S, Kim J-S, Kim M-S, Alhalaweh A, Cho W, Hwang S-J, Velaga SP. 2010. Bioavailability of indomethacin-saccharin cocrystals. J Pharm Pharmacol 62:1560–1568. 26. Noyes AA, Whitney WR. 1897. The rate of solution of solid substances in their own solutions. J Am Chem Soc 19:930–934. 27. Thommes M, Ely DR, Carvajal MT, Pinal R. 2011. Improvement of the dissolution rate of poorly soluble drugs by solid crystal suspensions. Mol Pharm 8:727–735. 28. Cheney ML, Weyna DR, Shan N, Hanna M, Wojtas L, Zaworotko MJ. 2011. Coformer selection in pharmaceutical cocrystal development: A case study of a meloxicam aspirin cocrystal that exhibits enhanced solubility and pharmacokinetics. J Pharm Sci 100:2172–2181.
Boksa, Otte, and Pinal, JOURNAL OF PHARMACEUTICAL SCIENCES
