Dissolution Enhancement in Cocoa Extract, Combining Hydrophilic Polymers through Hot-Melt Extrusion

The aim of this study was to improve the physicochemical properties of cocoa extract (CE) using hot-melt extrusion (HME) for pharmaceutical proposes. A mixture design was applied using three distinct hydrophilic polymeric matrices (Soluplus, Plasdone S630, and Eudragit E). Systems obtained by HME were evaluated using morphologic, chromatographic, thermic, spectroscopic, and diffractometric assays. The flow, wettability, and dissolution rate of HME powders were also assessed. Both CE and its marker theobromine proved to be stable under heating according to thermal analysis and Arrhenius plot under isothermal conditions. Physicochemical analysis confirmed the stability of CE HME preparations and provided evidence of drug–polymer interactions. Improvements in the functional characteristics of CE were observed after the extrusion process, particularly in dissolution and flow properties. In addition, the use of a mixture design allowed the identification of synergic effects by excipient combination. The optimized combination of polymers obtained considering four different aspects showed that a mixture of the Soluplus, Plasdone S630, and Eudragit E in equal proportions produced the best results (flowability index 88%; contact angle 47°; dispersibility 7.5%; and dissolution efficiency 87%), therefore making the pharmaceutical use of CE more feasible.


Introduction
Cocoa extract (CE; Theobroma cacao L.), one of the main Brazilian agricultural commodities, is largely composed of the flavonoid theobromine (TB) [1]. Besides its common use in the food industry, this natural product has exhibited diverse therapeutic potential, such as cardioprotective and anti-inflammatory actions [1]. Also, TB has shown a variety of possible pharmacological applications, including use as anti-carcinogenic or anticholesterolemic agents, and as a cough suppressant [2][3][4].
Vegetal sources have been widely explored therapeutically by traditional medicine; however, the insertion of new technologies in the development of this class of products is quite scarce, maintaining its "homemade" character. Furthermore, the poor mechanical and physicochemical properties of dried vegetal extracts can result in deficient flow behavior and inadequate compactness, hindering large-scale production [5]. Indeed, CE has remarkably deficient flow and wettability characteristics [6], while TB has restricted solubility and a poor dissolution profile [7,8].
Hot-melt extrusion (HME) has gained interest in the pharmaceutical field as a processing technology capable of producing solid dispersions with a high degree of drug-polymer interactions. The simultaneous mechanical and thermal shear of samples achieved by HME can noticeably modify Thermogravimetric tests were performed in CE samples as well as in the main marker TB in order to evaluate thermal stability. TGA analysis of TB using a constant heating rate revealed a monophasic drug decomposition in the range of 230-330 • C (T peak of first derivative = 305 • C), involving a complete sample weight loss. Decomposition of CE, meanwhile, occurred in a similar range of temperature (210-330 • C) as a well-defined decomposition first step with 43% weight loss (T peak of first derivative = 279 • C), likely corresponding to the decomposition of TB and other compounds found in CE such as polyphenols and organic matter [23]. TGA performed in inert (N 2 ) or oxidant atmosphere (synthetic air) showed the same profile, suggesting that oxygen did not affect the decomposition kinetics under the conditions studied.
The Arrhenius kinetic approach was applied based on TGA isothermal studies in an inert atmosphere. The activation energy values of TB and CE were 173.1 kJ mol −1 (r = 0.967) and 122.0 kJ mol −1 (r = 0.985), respectively, which are fairly high values when compared with those found for other drugs, which are usually within the range of 40-100 kJ mol −1 [24]. Indeed, TB activation energy is more than twice the value described for caffeine: 80.5 kJ mol −1 [25]. According to the Arrhenius plot, at the highest HME temperature used in this study (185 • C), TB and CE would take 163 min and 58 min, respectively, to lose 5% of their weight. Considering that the extrusion process takes less than 5 min, there are reasonable indications for the feasibility of thermal methods for CE processing without stability concerns.

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 • C, above at least 40 • C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%.

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the HME-PVP

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the

Physicochemical Characterization
Extrusion conditions were settled to obtain continuous extrusion, and processing temperatures were established at the range of 150-185 °C, above at least 40 °C of the glass transition of polymers for single matrix systems based on previous studies with the same polymers [8]. In HME-Sol, the rotation speed and temperature were adjusted to have a uniform strip with better flow. For systems with more than one polymer, the initial settled temperature was the lowest one used for the present polymers, but it had to be increased for HME-Sol-EuE and HME-Sol-PVP to obtain a more homogeneous extrudate with adequate extrusion flow. The filaments showed a brown color and a uniform appearance (Table 1). HPLC analyses revealed no decomposition signal with drug content in the range of 90-104%. The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the The morphological aspects of samples observed with optical microscopy and SEM ( Figure 1) showed that CE and the polymers could be clearly distinguished in PM, in contrast to the HME samples, which appeared homogeneous. Moreover, SEM micrographs revealed the dense structure of HME, as expected [12].
The thermal profile of mixtures was determined before (PM) and after the extrusion process (HME). TGA results showed no significant difference in thermal stability after processing. In fact, the extrudates presented a thermal profile equal to the sum of their individual compounds ( Figure 2) without evidence of drug-polymer incompatibility.
The crystallinity of the systems could not be measured by thermal techniques once the TB melting event occurred after its thermal degradation [8]. Therefore, for this kind of verification, XRPD analyses were performed in both PM and HME samples, as well as in CE and TB as supplied. The main Pharmaceutics 2018, 10, 135 6 of 14 characteristic peaks of crystalline TB were also found in CE at 13.38 • and 26.98 • 2θ, together with an amorphous component ( Figure 3).
XRPD of HME samples ( Figure 3) indicated a strong amorphous component in all systems influenced by the polymers and the CE (both amorphous). Despite the dilution effect, peaks of the crystalline state of TB were able to be identified in almost all samples, reinforcing the stability of the crystalline lattice of this marker and its consequent incorporation into a pharmaceutical matrix.
Pharmaceutics 2018, 10, x FOR PEER REVIEW 6 of 14 extrudates presented a thermal profile equal to the sum of their individual compounds ( Figure 2) without evidence of drug-polymer incompatibility. The crystallinity of the systems could not be measured by thermal techniques once the TB melting event occurred after its thermal degradation [8]. Therefore, for this kind of verification, XRPD analyses were performed in both PM and HME samples, as well as in CE and TB as supplied. The main characteristic peaks of crystalline TB were also found in CE at 13.38° and 26.98° 2θ, together with an amorphous component ( Figure 3).
XRPD of HME samples ( Figure 3) indicated a strong amorphous component in all systems influenced by the polymers and the CE (both amorphous). Despite the dilution effect, peaks of the crystalline state of TB were able to be identified in almost all samples, reinforcing the stability of the crystalline lattice of this marker and its consequent incorporation into a pharmaceutical matrix.   In FTIR spectra of CE, the main characteristic bands of TB were identified: in particular, the C=N stretching vibration band at 1689 cm −1 , the band related to C-N at 1222 cm −1 , and the bands associated with the two carbonyl group vibrations in the meta position at 1664 and 1544 cm −1 (Figure 4).  In FTIR spectra of CE, the main characteristic bands of TB were identified: in particular, the C=N stretching vibration band at 1689 cm −1 , the band related to C-N at 1222 cm −1 , and the bands associated with the two carbonyl group vibrations in the meta position at 1664 and 1544 cm −1 (Figure 4). In FTIR spectra of CE, the main characteristic bands of TB were identified: in particular, the C=N stretching vibration band at 1689 cm −1 , the band related to C-N at 1222 cm −1 , and the bands associated with the two carbonyl group vibrations in the meta position at 1664 and 1544 cm −1 (Figure 4). Nonetheless, small changes in the FTIR spectrum of HME samples mainly related to the intensity and position of some bands corresponding to the functional groups of TB suggest the interaction of this secondary metabolite with the polymeric matrix. Specifically, characteristic stretches with a slight shift or overlay were detected in HME-Sol, HME-Sol-EuE, and HME-Sol-EuE-PVP samples. In the HME PVP sample, C=N stretch and C-O were not visible. Moreover, the C=N stretches of HME Sol-PVP and PVP-EuE were not visible, reinforcing the probable drug-polymer interaction, possibly due to a higher degree of hydrogen bonding in the amorphous state [26].

Flow Evaluation
Flow behavior affects many industrial applications of pharmaceutical powders and is a crucial property in the development of oral solid dosage forms. Nevertheless, one of the main difficulties with bulk solid flow studies is the lack of reproducibility for most of the tests commonly used for this purpose, such as angle of repose and compressibility [27]. In this work, the assessment of multiple tests and the high degree of equipment automation allowed measures with a low variation coefficient (less than 3.5%), which resulted in a flowability index.
As expected, CE showed a low flowability (=49), proving its poor flow capacity (Table 2). Meanwhile, HME extrudates showed flowability in the range of 71-88, demonstrating a remarkable improvement of this property, which enables this processed natural product to be used in direct compression of tablets or to fill capsules, without any additional processing [28].
The thermal shear caused by HME is not only capable of greatly increasing the degree of interaction between the drug and polymer but also gives rise to a dense and uniform matrix producing granules with excellent flow characteristics, as previously described [9,11].
Together with the flowability determination, the dispersibility (%), which measures the tendency of a powder to scatter in the air, was also assessed. Once again, there was a great difference in performance between CE (37.8%) and HME extrudates (below 11%), as described in Table 2. A substantial improvement was achieved after HME processing, mitigating the possibility of air contamination in the production area. Nonetheless, small changes in the FTIR spectrum of HME samples mainly related to the intensity and position of some bands corresponding to the functional groups of TB suggest the interaction of this secondary metabolite with the polymeric matrix. Specifically, characteristic stretches with a slight shift or overlay were detected in HME-Sol, HME-Sol-EuE, and HME-Sol-EuE-PVP samples. In the HME PVP sample, C=N stretch and C-O were not visible. Moreover, the C=N stretches of HME Sol-PVP and PVP-EuE were not visible, reinforcing the probable drug-polymer interaction, possibly due to a higher degree of hydrogen bonding in the amorphous state [26].

Flow Evaluation
Flow behavior affects many industrial applications of pharmaceutical powders and is a crucial property in the development of oral solid dosage forms. Nevertheless, one of the main difficulties with bulk solid flow studies is the lack of reproducibility for most of the tests commonly used for this purpose, such as angle of repose and compressibility [27]. In this work, the assessment of multiple tests and the high degree of equipment automation allowed measures with a low variation coefficient (less than 3.5%), which resulted in a flowability index.
As expected, CE showed a low flowability (=49), proving its poor flow capacity (Table 2). Meanwhile, HME extrudates showed flowability in the range of 71-88, demonstrating a remarkable improvement of this property, which enables this processed natural product to be used in direct compression of tablets or to fill capsules, without any additional processing [28].
The thermal shear caused by HME is not only capable of greatly increasing the degree of interaction between the drug and polymer but also gives rise to a dense and uniform matrix producing granules with excellent flow characteristics, as previously described [9,11].
Together with the flowability determination, the dispersibility (%), which measures the tendency of a powder to scatter in the air, was also assessed. Once again, there was a great difference in performance between CE (37.8%) and HME extrudates (below 11%), as described in Table 2. A substantial improvement was achieved after HME processing, mitigating the possibility of air contamination in the production area. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. 20 Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. 26 Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. 18 Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination.

± 2.6
Pharmaceutics 2018, 10, x FOR PEER REVIEW 9 of 14 Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. 20 Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination.
15.9 ± 3.3 88.0 ± 1.6 7.5 ± 1.0 The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination. Although considerable improvements in flowability and dispersibility have been reached through HME processing, differences in these properties can be observed according to the composition of the polymer matrix, as shown in the response surfaces in Figure 5. The predictive equation for flowability was adjusted to a special cubic model (p < 0.001) with r 2 = 0.963 ( Figure 5A). The three polymers used in this study, each contribute in equal terms to the flow improvement, showing high coefficient values with a positive signal ( Figure 5A). The important contribution of the triple interaction between the polymers (high coefficient value) should also be highlighted, indicating a synergism among the different materials ( Figure 5A).
The predictive equation for dispersibility was calculated using a linear method (p < 0.0001, r 2 = 0.735; Figure 5B). In this case, there was no interaction between the formulation components. However, there are differences in polymer performance, especially for Sol and PVP, which may generate denser granules with lower potential for environmental contamination.

Wettability
Wettability is the ability of a solid surface to overcome the cohesive intermolecular interactions of water, mainly from hydrogen bonds, and to allow liquid spreading. This property is a sine qua non condition for the dissolution of solid dosage forms administered orally [29]. According to a previous report, CE presents a deficient wettability [7]. Figure 6A displays the contact angle results calculated for the seven HME extrudates. All samples presented a contact angle smaller than 90°, which indicates a favorable wetting of the extrudate surface. Similar results involving a hydrophilic HME matrix are described by other authors, due to the strong solid-liquid interaction achieved by these systems [30].

Wettability
Wettability is the ability of a solid surface to overcome the cohesive intermolecular interactions of water, mainly from hydrogen bonds, and to allow liquid spreading. This property is a sine qua non condition for the dissolution of solid dosage forms administered orally [29]. According to a previous report, CE presents a deficient wettability [7]. Figure 6A displays the contact angle results calculated for the seven HME extrudates. All samples presented a contact angle smaller than 90 • , which indicates a favorable wetting of the extrudate surface. Similar results involving a hydrophilic HME matrix are described by other authors, due to the strong solid-liquid interaction achieved by these systems [30].

Wettability
Wettability is the ability of a solid surface to overcome the cohesive intermolecular interactions of water, mainly from hydrogen bonds, and to allow liquid spreading. This property is a sine qua non condition for the dissolution of solid dosage forms administered orally [29]. According to a previous report, CE presents a deficient wettability [7]. Figure 6A displays the contact angle results calculated for the seven HME extrudates. All samples presented a contact angle smaller than 90°, which indicates a favorable wetting of the extrudate surface. Similar results involving a hydrophilic HME matrix are described by other authors, due to the strong solid-liquid interaction achieved by these systems [30].  Although all formulations showed good wettability, marked differences could be observed between them, which indicate that the polymer composition plays an important role in this property. In fact, the response surface ( Figure 6B) built from a quadratic model (p < 0.001; r 2 = 0.940) reveals that Sol presents the lowest coefficient of the predictive equation, therefore leading to lower contact angles. Moreover, the combination of polymers favors the reduction of the contact angle, especially in the regions of the contour plot that combine PVP and EuE (significant interaction between these two polymers with a negative signal). The amphiphilic properties of Sol and the wettability enhancing activity attributed to PVP and EuE as described in the literature corroborate our results [31].

Dissolution Rate
Dissolution profiles of CE as supplied, compared with those of CE extrudates, are shown in Figure 7A. In contrast with the dissolution described for TB (CE marker), which shows a slow dissolution behavior [8], CE itself as well as HME formulations practically reached their maximum TB dissolution level in the initial minutes of the experiment. This behavior can be attributed to other components of the CE that might work as solubilizers. However, differences between the maximum solubilization levels were noticed between samples. While CE dissolves only 75% of the dose, the extrudates dissolved up to 95% of their doses, as in the case of HME Sol-EuE and HME Sol-EuE-PVP. Indeed, statistical analysis showed these two systems had better performance when compared with CE, with DE30 of 84 and 87%, respectively, against 65% as found for CE ( Figure 7A).
As observed in other assays, the composition of the formulation has an important influence on the dissolution of HME samples. The quadratic model applied to this response (p = 0.011; r 2 = 0.597) shows that regions composed of mixtures containing predominantly Sol and EuE led to DE30 results above 85%. In fact, the coefficient for the Sol-EuE interaction presented a high value, strongly contributing to the predictive equation ( Figure 7B). Although all formulations showed good wettability, marked differences could be observed between them, which indicate that the polymer composition plays an important role in this property. In fact, the response surface ( Figure 6B) built from a quadratic model (p < 0.001; r 2 = 0.940) reveals that Sol presents the lowest coefficient of the predictive equation, therefore leading to lower contact angles. Moreover, the combination of polymers favors the reduction of the contact angle, especially in the regions of the contour plot that combine PVP and EuE (significant interaction between these two polymers with a negative signal). The amphiphilic properties of Sol and the wettability enhancing activity attributed to PVP and EuE as described in the literature corroborate our results [31].

Dissolution Rate
Dissolution profiles of CE as supplied, compared with those of CE extrudates, are shown in Figure 7A. In contrast with the dissolution described for TB (CE marker), which shows a slow dissolution behavior [8], CE itself as well as HME formulations practically reached their maximum TB dissolution level in the initial minutes of the experiment. This behavior can be attributed to other components of the CE that might work as solubilizers. However, differences between the maximum solubilization levels were noticed between samples. While CE dissolves only 75% of the dose, the extrudates dissolved up to 95% of their doses, as in the case of HME Sol-EuE and HME Sol-EuE-PVP. Indeed, statistical analysis showed these two systems had better performance when compared with CE, with DE30 of 84 and 87%, respectively, against 65% as found for CE ( Figure 7A).
As observed in other assays, the composition of the formulation has an important influence on the dissolution of HME samples. The quadratic model applied to this response (p = 0.011; r 2 = 0.597) shows that regions composed of mixtures containing predominantly Sol and EuE led to DE30 results above 85%. In fact, the coefficient for the Sol-EuE interaction presented a high value, strongly contributing to the predictive equation ( Figure 7B).

Prediction of the Optimized Formulation
Mixture designs enable the determination, within the range of the excipient concentration studied, of regions in which different evaluated responses can be considered simultaneously [12]. In the case of this study, we chose to obtain an optimized response considering the four aspects of the study: DE30, flowability, contact angle, and dispersibility ( Figure 8).
The highest desirability index (0.75) was found with a polymeric mixture composed of 23.1% Sol, 23.8% PVP, and 23.1% EuE, which practically corresponds to the central region of the contour plot. The expected responses for this formulation were DE30 of 84%, flowability of 87%, dispersibility of 8.4%, and contact angle of 48°, which are nearly the same as the results determined experimentally for the formulation HME Sol-PVP-EuE.
Moreover, a large central region, whose composition leads to desirability greater than 0.7, is observed in the contour plot (gray area of contour diagram). The possibilities with the best performance include a formulation containing 45.5% Sol, 11.9% PVP, and 12.6% EuE and a formulation containing 4.9% Sol, 32.2% PVP, and 33.6% EuE (Figure 8).

Prediction of the Optimized Formulation
Mixture designs enable the determination, within the range of the excipient concentration studied, of regions in which different evaluated responses can be considered simultaneously [12]. In the case of this study, we chose to obtain an optimized response considering the four aspects of the study: DE30, flowability, contact angle, and dispersibility ( Figure 8).
The highest desirability index (0.75) was found with a polymeric mixture composed of 23.1% Sol, 23.8% PVP, and 23.1% EuE, which practically corresponds to the central region of the contour plot. The expected responses for this formulation were DE30 of 84%, flowability of 87%, dispersibility of 8.4%, and contact angle of 48 • , which are nearly the same as the results determined experimentally for the formulation HME Sol-PVP-EuE.
Moreover, a large central region, whose composition leads to desirability greater than 0.7, is observed in the contour plot (gray area of contour diagram). The possibilities with the best performance include a formulation containing 45.5% Sol, 11.9% PVP, and 12.6% EuE and a formulation containing 4.9% Sol, 32.2% PVP, and 33.6% EuE (Figure 8).

Prediction of the Optimized Formulation
Mixture designs enable the determination, within the range of the excipient concentration studied, of regions in which different evaluated responses can be considered simultaneously [12]. In the case of this study, we chose to obtain an optimized response considering the four aspects of the study: DE30, flowability, contact angle, and dispersibility ( Figure 8).
The highest desirability index (0.75) was found with a polymeric mixture composed of 23.1% Sol, 23.8% PVP, and 23.1% EuE, which practically corresponds to the central region of the contour plot. The expected responses for this formulation were DE30 of 84%, flowability of 87%, dispersibility of 8.4%, and contact angle of 48°, which are nearly the same as the results determined experimentally for the formulation HME Sol-PVP-EuE.
Moreover, a large central region, whose composition leads to desirability greater than 0.7, is observed in the contour plot (gray area of contour diagram). The possibilities with the best performance include a formulation containing 45.5% Sol, 11.9% PVP, and 12.6% EuE and a formulation containing 4.9% Sol, 32.2% PVP, and 33.6% EuE (Figure 8).

Conclusions
Solid dispersions of CE were effectively produced using a single-step HME process. Natural extract stability was preserved while dissolution and flow properties were improved. Despite the high stability of the TB crystal lattice, there was physicochemical evidence of drug-polymer interactions. The use of a mixture design allowed the identification of synergistic effects with excipient combination. In fact, the optimized formulation obtained considering four different responses showed that a mixture of Sol, PVP, and EuE in equal proportions produced the best results, making the pharmaceutical use of CE more feasible.