Native starch as in situ binder for continuous twin screw wet granulation

Native starch as in situ binder for continuous twin screw wet granulation

International Journal of Pharmaceutics 571 (2019) 118760 Contents lists available at ScienceDirect International Journal of Pharmaceutics journal ho...

3MB Sizes 1 Downloads 17 Views

International Journal of Pharmaceutics 571 (2019) 118760

Contents lists available at ScienceDirect

International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm

Native starch as in situ binder for continuous twin screw wet granulation a

a

a

b

T

b

Lise Vandevivere , Christoph Portier , Valérie Vanhoorne , Olaf Häusler , Denis Simon , ⁎ Thomas De Beerc, Chris Vervaetd, a

Ghent University, Laboratory of Pharmaceutical Technology, Ottergemsesteenweg 460, 9000 Gent, Belgium Roquette Frères, Rue de la Haute Loge, 62136 Lestrem, France Ghent University, Laboratory of Pharmaceutical Process Analytical Technology, Ottergemsesesteenweg 460, 9000 Ghent, Belgium d Ghent University, Laboratory of Pharmaceutical Technology, Ottergemsesteenweg 460, 9000 Gent, Belgium b c

A R T I C LE I N FO

A B S T R A C T

Keywords: Continuous wet granulation Twin screw granulator Native starch Gelatinization Binder Granule properties Continuous manufacturing

The use of native starch as in situ binder in a continuous twin screw wet granulation process was studied. Gelatinization of pea starch occurred in the barrel of the granulator using a poorly soluble excipient (anhydrous dicalcium phosphate), but the degree of gelatinization depended on the liquid-to-solid ratio, the granule heating and the screw configuration. Furthermore, the degree of starch gelatinization was correlated with the granule quality: higher binder efficiency was observed in runs where starch was more gelatinized. SEM and PLOM images showed experimental runs which resulted in completely gelatinized starch. Other starch types (maize, potato and wheat starch) could also be gelatinized when processed above a critical barrel temperature for gelatinization. This barrel temperature was different for all starches. In situ starch gelatinization was also investigated in combination with a highly soluble excipient (mannitol). The lower granule friability observed using pure mannitol compared to a mannitol/starch mixture indicated that starch did not contribute to the binding, hence starch did not gelatinize during processing. The study showed that native starch can be considered as a promising in situ binder for continuous twin screw wet granulation of a poorly soluble formulation.

1. Introduction An optimal continuous granulation process integrates the continuous mixing and agglomeration of powders, which can immediately be used for downstream processing (Vervaet and Remon, 2005). Therefore, continuous twin screw wet granulation (TSG) is of great interest as it can be implemented into a fully continuous from-powderto-tablet line. As a switch to continuous manufacturing is quickly gaining momentum in the pharmaceutical industry, the suitability of excipients needs to be evaluated towards this manner of processing. Continuous wet granulation is a complex process in which several critical formulation and process variables affect the outcome (Hoag, 2014). While much research has already been performed on the influence of process parameters on granule quality, less studies have assessed the effects of the formulation (Keleb et al., 2004; Keleb et al., 2002; Van et al., 2008; Djuric et al., 2009; Vanhoorne et al., 2016; Saleh et al., 2015). A binder is often added to the formulation to facilitate the granulation process (e.g. powder wetting and consolidation of the agglomerates). As the binder decisively affects the mechanical properties of granules and tablets, the selection of a binder during TSG to

efficiently agglomerate the active pharmaceutical ingredient with excipients is critical. The functionality of the binder depends on its intrinsic binder capacity and the distribution of the binder in the powder bed. When the binder is added as a dry powder to the powder blend, it is essential that the binder interacts with the granulation fluid added during the granulation process. However, the short residence time (5 – 20 s) and the limited amount of liquid used in TSG are both challenges in order to hydrate the binder and activate its binding properties (Saleh et al., 2015; El et al., 2013). The most common pharmaceutical binders are sugars (e.g. maltodextrins), natural polymers (e.g. starch), synthetic and semi-synthetic polymers (e.g. polyvinylpyrrolidone) and cellulosebased polymers (e.g. hydroxypropyl methylcellulose and hydroxypropylcellulose). Native starches are natural non-toxic polymers which are inexpensive and compatible with most APIs. Nevertheless, its low cold water solubility has limited its application (Hoag, 2014). Starch granules consist almost entirely of two polysaccharides, amylose and amylopectin. Amylose is mainly a linear polymer, whereas the amylopectin molecule is much larger and branched. Starch granules are semi-crystalline as these contain crystalline and amorphous parts. When heated in an aqueous medium, the starch granule is initially



Corresponding author. E-mail addresses: [email protected] (L. Vandevivere), [email protected] (C. Portier), [email protected] (V. Vanhoorne), [email protected] (O. Häusler), [email protected] (D. Simon), [email protected] (T. De Beer), [email protected] (C. Vervaet). https://doi.org/10.1016/j.ijpharm.2019.118760 Received 4 August 2019; Received in revised form 30 September 2019; Accepted 1 October 2019 Available online 14 October 2019 0378-5173/ © 2019 Elsevier B.V. All rights reserved.

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 1. Schematic overview of the screw configurations with conveying elements (green) and kneading elements (orange) with an L/D of ¼. The horizontal arrow indicates the powder flow direction. The kneading elements highlighted in blue represents the location of the two kneading elements with a 90° stagger angle (if applicable). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

system, GEA Pharma Systems, Wommelgem, Belgium). This continuous granulator consists of two co-rotating screws with a length-to-diameter (L/D) of 20:1. Five formulations were obtained by preblending DCP (95% w/w) with the different native starches (5% w/w) and one by preblending mannitol (95% w/w) with pea starch (5% w/w). These preblends were prepared in a tumbling mixer (Inversina Bioengineering, Wald, Switzerland) for 15 min at 25 rotations per minute (rpm). Each dry premix was transferred to the twin-screw lossin-weight feeder (KT20, K-Tron Soder, Niederlenz, Switzerland) and subsequently dosed into the barrel of the granulator. The granulation liquid was gravimetrically pumped into the barrel via dual liquid feed ports, injecting the liquid on top of each screw. This addition was performed using two out-of-phase peristaltic pumps (Watson Marlow, Cornwall, UK) and two silicon tubings (internal and external diameter of 1.6 mm) both connected to a 1.6 mm nozzle. The barrel jacket was equipped with an active cooling system in order to maintain the set temperature during processing. Torque was monitored by a built-in torque-gauge at 1-second intervals. After stabilization of the torque, 800 g of the produced granules were sampled at the outlet of the granulator and tray dried at 40 °C until a loss-on-drying value between 1 and 3% was achieved.

hydrated in the amorphous part and starts to swell. A further increase of temperature results in destabilization of the crystalline part. Eventually, the swelling becomes irreversible and the starch structure fully disintegrates as the intermolecular hydrogen bonds which maintain the structural integrity of the starch granule are destroyed. This process is called gelatinization and the temperature at which gelatinization occurs is defined as the gelatinization temperature. The degree of gelatinization correlates with the amount of available water and heating. Since both parameters can be varied during TSG, this technique can possibly induce in situ gelatinization of starch during TSG in order to activate the binding properties of starch added as a dry powder to the formulation (Hoag, 2014; AI PRESS; Lund and Lorenz, 1984; Remon et al., 1990). The potential use of starches as binding agent has already been investigated in a conventional batch process by Visavarungroj et al. (Remon et al., 1990). Furthermore, starches which were partially pregelatinized are useful both as binder and disintegrant in formulations processed with hear shear granulation (Rahman et al., 2008). It was our aim to investigate the use of native starch as binder in a continuous wet granulation process, and to analyze if gelatinization of starch took place in the barrel of the twin screw granulator and how process and formulation parameters affected the degree of gelatinization. Furthermore, the quality of produced granules was evaluated via an extensive granule characterization to determine the correlation between the degree of starch gelatinization and the granule quality. Pea starch was selected as model, as it is considered to gelatinize very easily. Other native starches (maize, potato and wheat starch) were also evaluated as possible binder.

2.2.2. Evaluation of a poorly soluble formulation 2.2.2.1. Design of experiments. Two Designs of Experiments (DoE) were executed. Both included DCP and pea starch (95:5) as formulation. Analysis was performed using MODDE 12.0 software (Sartorius Stedim Biotech, Umeå, Sweden). 2.2.2.1.1. First screening design. A fractional factorial design (resolution V+) with 20 experiments was used to evaluate the influence of four process parameters on the granulation process and the granule properties: L/S-ratio (0.10 – 0.30), screw speed (300 – 900 rpm), barrel temperature (30 – 60 °C) and throughput (10 – 20 kg/ h). Additionally, two different screw configurations were studied: (i) one configuration containing 1 kneading zone of 4 kneading elements (1 × 4) and (ii) a second configuration containing 2 kneading zones of 6 kneading elements separated by a conveying zone (1.5 L/D) (2 × 6). The kneading elements of both configurations were arranged at a forward stagger angle of 60°. An overview of the screw configurations is shown in Fig. 1. The center points with screw configurations 1 × 4 and 2 × 6 were run in duplicate. The different factor settings for each run are listed in Table 1. 2.2.2.1.2. Second screening design. Based on preliminary results and on results obtained from the first DoE, a 2-level full-factorial DoE was performed. The influence of three factors was evaluated on the granulation process and the granule properties: L/S-ratio (0.18 – 0.26), screw speed (300 – 400 rpm) and the barrel temperature (45 –

2. Materials and methods 2.1. Materials Anhydrous dicalcium phosphate (DCP) (Calipharm® A, obtained from Innophos, Chicago Heights, USA) and mannitol (Pearlitol® 50 C, donated by Roquette Frères, Lestrem, France) were used as poorly soluble and highly soluble model excipients, respectively. Native starches (pea, maize, potato and wheat starch) were kindly donated by Roquette Frères (Lestrem, France) and were used as pharmaceutical in situ binders. All tested starches were of pharmaceutical grade. Demineralized water was added as granulation liquid. 2.2. Granulation experiments 2.2.1. Preparation of the granules Granulation experiments were performed using the twin screw granulator of a continuous from-powder-to-tablet line (ConsiGmaTM-25 2

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

2.2.3. Evaluation of a highly soluble formulation Pure mannitol as well as a dry preblend of mannitol and pea starch (95:5) were granulated at constant process parameters (screw speed of 500 rpm and a throughput of 20 kg/h) using a fixed screw configuration (Fig. 1): 2 kneading zones of 6 kneading elements (2 × 6) with the last two kneading elements of each zone set at a forward stagger angle of 90°, while the residual kneading elements had a stagger angle of 60°. The two kneading zones were separated by a conveying zone (1.5 L/D). These parameters were also selected based on results obtained from the two screening designs. Due to the lower density of mannitol, a higher screw speed (500 rpm) was used compared to the DCP formulation to allow processing a 20 kg/h throughput without overfilling of the granulation barrel. The granule quality was determined at two barrel temperatures (55 and 65 °C). The L/S-ratio was varied between 0.08 and 0.1425 on four levels.

Table 1 Overview of the experimental screening design. Run

L/S-ratio

Screw speed (rpm)

Barrel temperature (°C)

Throughput (kg/h)

Screw configuration

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2

300 900 300 900 300 900 300 900 300 900 300 900 300 900 300 900 600 600 600 600

60 30 30 60 30 60 60 30 30 60 60 30 60 30 30 60 45 45 45 45

10 10 10 10 20 20 20 20 10 10 10 10 20 20 20 20 15 15 15 15

1×4 1×4 2×6 2×6 1×4 1×4 2×6 2×6 1×4 1×4 2×6 2×6 1×4 1×4 2×6 2×6 1×4 1×4 2×6 2×6

2.3. Characterization methods 2.3.1. Granule characterization 2.3.1.1. Particle size analysis. Granule size was analyzed via dynamic image analysis using the QICPIC™ system (Sympatec, ClausthalZellerfeld, Germany) equipped with a vibrating feeder system (Vibri/ L™) for gravimetrical feeding of the granules. Each sample was measured five times. Windox 5 software (Sympatec, ClausthalZellerfeld, Germany) was used to calculate the median granule size (d50) as the equivalent projected circle diameter based on a volume distribution. The fraction of fines and oversized granules were defined as < 150 µm and > 850 µm, respectively.

Table 2 Overview of the experimental screening design. Run

L/S-ratio

Screw speed (rpm)

Barrel temperature (°C)

1 2 3 4 5 6 7 8 9 10 11

0.18 0.18 0.26 0.26 0.18 0.18 0.26 0.26 0.22 0.22 0.22

300 400 300 400 300 400 300 400 350 350 350

45 45 45 45 60 60 60 60 52.5 52.5 52.5

2.3.1.2. Friability testing. The granule friability was determined in triplicate using a friabilator (PTFE Pharma Test, Hainburg, Germany) at a speed of 25 rpm for 10 min, by subjecting 10 g (Iwt) of granules together with 200 glass beads (mean diameter 4 mm) to falling shocks. Prior to determination, the granule fraction < 250 µm was removed. Afterwards, the glass beads were removed and the mass retained on a 250 µm sieve (Fwt) was determined. The friability was calculated as [(Iwt – Fwt)/Iwt] * 100. 2.3.1.3. Polarized light and scanning electron microscopy. To determine the birefringence pattern of native starch, samples were prepared by dispersing 10 g of granules in 50 mL demineralized water. These samples were analyzed with polarized light optical microscopy (PLOM) using a Leica DM2500 P microscope (Leica Microsystems, Diegem, Belgium) at 40× magnification. Furthermore, a DMRB Leitz polarized light optical microscope (Leica Microsystems, Diegem, Belgium) equipped with a lambda plate (PLOM-LP) at 140x and 70x magnification was used to obtain colored microscopic images. With PLOM-LP, polarization and (to a lesser extent) shape can be evaluated. To study the degree of gelatinization, images were recorded with a Quanta 200 FEG scanning electron microscope (SEM) (Thermo Fisher Scientific, Waltham, Massachusetts, USA). With this method, detailed observations focusing on the size and shape of the starch granules can be obtained. For this purpose, a small amount of the product was deposited on an aluminum stub covered with conductive carbon tape. The used electron accelerating voltage was 12.5 kV. Complete gelatinization of starch is characterized by a total disruption of its native structure and due to the loss of polarization, no birefringence is observed.

60 °C). Three replicates of the center point were run. The different factor settings for each experiment are listed in Table 2. The throughput was fixed at 20 kg/h, and the screw configuration consisted of 2 kneading zones of 6 kneading elements (2 × 6) separated by a conveying zone (1.5 L/D). Moreover, the last two kneading elements of each zone were set at a forward stagger angle of 90°, while the residual kneading elements had a stagger angle of 60°. This configuration prolonged the residence time because of the higher flow restriction for the material (Fig. 1).

2.2.2.2. Comparison between different types of native starches. Granulation of DCP/starch blends was performed at constant process parameters which were based on results obtained from the screening designs: a screw speed of 300 rpm and a throughput of 20 kg/ h. A fixed screw configuration was used: 2 kneading zones of 6 kneading elements (2 × 6) with the last two kneading elements of each zone set at a forward stagger angle of 90°, while the residual kneading elements had a stagger angle of 60°. The two kneading zones were separated by a conveying zone (1.5 L/D) (Fig. 1). The L/S-ratio was varied between 0.180 and 0.255 on five levels. Based on preliminary studies, two barrel temperatures for each formulation were studied. The first was similar for all formulations and was set at 50 °C. The second temperature (the TSG gelatinization temperature) depended on the starch type and was determined by the lowest possible temperature where the native starch enabled efficient granulation.

2.3.2. Native starch characterization The native starches were characterized by measuring the gelatinization temperature of each starch with differential scanning calorimetry (DSC). A DSC 8000 (PerkinElmer, Waltham, Massachusetts, USA) was used to perform the thermal analysis of the native starches. Samples consisted of 18% m/m of native starch dispersed in demineralized water. Each sample was first equilibrated at 5 °C, followed by a 3

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 2. Microscopic images showing (a) Maltese crosses (run 7) and (b) disruption of the crystalline structure (no Maltese crosses were detected) (run 11). The Maltese crosses are indicated in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

generation by friction (Stauffer et al., 2019). Consequently, the lack of Maltese cross in the granules manufactured via run 15 was likely linked to granule heating caused by friction, which induced the onset of gelatinization of the native starch.

heating run at 10 °C/min to 110 °C, which is above the gelatinization temperatures of all native starches. 3. Results and discussion 3.1. Starch as in situ binder – Evaluation of a poorly soluble formulation

3.1.1.2. Effect of the in situ starch gelatinization on granule properties. It was shown that in order to obtain in situ starch gelatinization, a combination of two factors was needed: an excess of liquid and sufficient granule heating. It was of interest to evaluate how this in situ gelatinization affected the granule quality. Therefore, the particle size and granule friability were determined for all samples. A significant relationship between the L/S-ratio and the particle size distribution of granules was detected: increasing the L/S-ratio yielded fewer fines, more oversized particles and a larger d50 (Fig. 3). At high L/ S-ratio, more liquid was available to obtain gelatinization of starch. This gelatinization activated the binding properties of starch, resulting in granule growth. The granule growth can be fully attributed to the achieved binding properties of starch as no strong interactions between DCP and the granulation liquid can take place due to the high water insolubility of DCP. As the L/S-ratio was the only significant main factor, solely interactions which included the L/S-ratio will be discussed further. An interaction between screw configuration and L/S-ratio influenced d50, fines and oversized fraction (Fig. 3). At low L/S-ratio, a higher fines fraction was obtained with a high number of kneading elements (2 × 6). However, at higher L/S-ratio the opposite effect was observed as more fines were yielded with a low number of kneading elements (1 × 4). At low L/S-ratio, insufficient granulation liquid was available to adequately wet the particles which limited the gelatinization of starch and consequently, no binding properties were achieved. As these fragile agglomerates were not resistant to the intense densification in case of a high number of kneading elements, this resulted in a larger fines fraction. In contrast, when sufficient granulation liquid was available, stronger bonds between the particles were formed due to a higher degree of starch gelatinization. This allowed granule densification and growth using a screw with a higher number of kneading elements. A similar effect was observed for the oversized fraction. The interaction between the L/S-ratio and the temperature was also significant with respect to fines fraction and d50. A similar effect for both responses was obtained. At high L/S-ratio, a higher temperature yielded a higher d50 compared to processing at low temperature. As a higher barrel temperature caused more granule heating, the degree of starch gelatinization increased. This may have resulted in the higher d50. At low L/S-ratio, no effect on d50 was found in function of temperature. This can be explained by the limited availability of liquid, which prevented starch gelatinization. As DCP is poorly water soluble, no significant impact of temperature on the properties of the excipient was assumed.

Two Designs of Experiments (DoE) were executed with DCP as model excipient. Both designs included pea starch, which is considered to gelatinize more easily compared to certain other starches. DSC experiments with pea starch indicated a gelatinization temperature of 57.5 (onset) − 64.2 (peak temperature) °C. 3.1.1. First screening design 3.1.1.1. In situ gelatinization of native starch. A Maltese cross appears when a native starch granule is examined under polarized light with an optical microscope. This birefringence pattern occurs because of the radially ordered arrangement of the crystalline areas in native starch. When the gelatinization occurs, the crystalline structure of starch is disrupted, leading to the disappearance of the Maltese cross (Fig. 2) (Bertoft, 2017; Bogracheva et al., 1998). Only three experimental runs (runs 11, 15 and 16) showed a disruption of the crystalline structure as no Maltese crosses were seen. Consequently, for these runs, in situ gelatinization of starch had occurred. Granules of runs 11, 15 and 16 were all produced at high L/S-ratio. This indicated that in order to gelatinize starch, a high liquid amount was required. However, to successfully obtain starch gelatinization, the granule temperature must be above the gelatinization temperature (Lund and Lorenz, 1984). As these three runs showed no Maltese cross, the combination of the factor settings must have resulted in sufficient granule heating. The sufficient granule heating during these three runs can be linked to different causes: for run 11 and 16, the high barrel temperature affected the granule heating. In contrast, run 15 was processed with a barrel temperature at low level, but the lack of Maltese crosses was correlated to the combination of factor settings: a high number of kneading elements and a high filling degree (due to a low screw speed combined with high throughput). Vercruysse et al. related the number of kneading elements to friction as the author reported a higher temperature increase at the barrel wall as a result of heat generated by friction when more kneading elements were used. More friction inside the barrel was generated when the mass flow through the barrel was restricted (Vercruysse et al., 2012). Moreover, by studying the influence of different fill levels on the granulation process, Meier et al. demonstrated that a higher filling degree resulted in an increased temperature (Meier et al., 2017). The higher shear generated during run 15 was also reflected in its torque value, the highest of all experimental runs (6.5 Nm). Stauffer et al. correlated an increase in granule temperature with a higher torque reading as torque was related with heat 4

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 3. Effect plot showing the influence of the granulation parameters on (a) fines fraction, (b) oversized fraction and (c) d50.

SEM images of run 7 demonstrated that pea starch maintained its original size (30 µm) as no gelatinization had occurred (Fig. 4a). As the structure of the starch granule was still preserved, no loss of polarization was observed with PLOM-LP. The run resulted in a considerably higher fines and smaller oversized fraction. A higher degree of starch gelatinization was observed, in ascending order, for runs 15, 16 and 11. In run 15, the gelatinization process was only initiated during granulation as the starch structure was mainly intact. The PLOM-LP images detected a high percentage of birefringent starch granules, indicative a low degree of gelatinization. At these settings starch was also separated from DCP, without significant interactions between both fractions (Fig. 4c). SEM images of run 16 showed more interactions between starch and DCP as starch better enclosed the DCP particles (Fig. 4d), which resulted in significantly less fines and more oversized granules (Table 3). SEM images of granules manufactured via run 11 showed mainly small and highly deformed starch granules, with extensive gelatinization of the starch granules (Fig. 4b). However, complete gelatinization was not obtained at these settings as some starch granules still showed birefringence. Run 11 resulted in the highest degree of gelatinization of all experimental runs and consequently, in granules of

Furthermore, the friability of granules produced at low L/S-ratio, could not be determined due to the extremely low particle fraction below 250 µm (i.e. size of sieve). Therefore, only the friability of granules obtained via runs with an L/S-ratio of 0.3 was evaluated. Most experimental runs resulted in granules with a friability above 85%, indicating limited binding strength for these granules. Only run 11 resulted in a low granule friability (19.3%) (Table 3), which also showed the lowest fines (1.6%) and highest oversized fraction (61.9%). These granules were all produced at high L/S-ratio and with high number of kneading elements (similar to runs 15 and 16 which also combined a low fines and high oversized fraction). These runs (runs 11, 15 and 16) also showed no Maltese cross, indicating that gelatinization of starch clearly affected the granule properties. As different granule quality between the different runs was observed, two microscopic techniques (SEM and PLOM-LP) were used for a thorough analysis of the degree of starch gelatinization. Four experimental runs (7, 11, 15 and 16) were evaluated with both microscopic methods. Run 7 was observed because of its extremely high fines fraction and low percentage of oversized particles. The microscopic images are illustrated in Fig. 4.

5

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

with the degree of gelatinization, a higher fines and lower oversized fraction as well as a higher granule friability was observed for these runs (Fig. 5). Runs 5 until 8 were performed at high barrel temperature, with only run 6 showing the Maltese cross. The latter only differed from run 5 in its screw speed, the lower screw speed of run 5 yielded a higher filling degree for run 5. This indicated that a higher filling degree was essential to induce sufficient granule heating in order to gelatinize starch when the L/S-ratio was set at its low level (0.18). This indicated that even at low L/S-level, already sufficient liquid was available to obtain starch gelatinization. The filling degree of run 6 was similar to run 8 but a higher L/S-ratio was used for granulation of run 8, showing no Maltese cross. As already sufficient liquid was available to result in gelatinization at an L/S-ratio of 0.18, the higher amount of liquid with run 8 may have contributed to the granule heating. As a higher amount of liquid resulted in a higher wetting degree of the mass in the granulator barrel, the residence time was prolonged and hence, more intense densification (Dhenge et al., 2010). When comparing the runs 5 to 8, the runs where starch was gelatinized showed again a higher binder efficiency (less fines, more oversized granules, lower granule friability) (Fig. 5). As run 7 showed good granule quality, it was analyzed with SEM and PLOM-LP. This run combined the filling degree, the L/S-ratio and the barrel temperature at their highest and high degree of interaction between DCP particles and starch was found (Fig. 6a). Moreover, the PLOM-LP images showed a total loss of starch polarization and no particles were identified with the original or swollen shape of native pea starch (Fig. 6b). According to these microscopic results, it can be concluded that the gelatinization process of native pea starch with run 7 was completed as all the native starch was gelatinized. This again showed a strong correlation between the granule quality and the degree of starch gelatinization. Furthermore, runs 5, 7 and 8 resulted in a low granule friability (12.3, 1.0 and 7.2%, respectively). Stauffer et al. showed that the main sources of heat transfer occurring during TSG were enthalpy of wetting and friction forces (both inducing an increase of the granule temperature). Therefore, densification (which is creating friction) results in granule heating Stauffer et al. (2019). As a higher degree of densification was obtained with runs 5, 7 and 8, granule heating was more intense, which positively affected the degree of starch gelatinization. This indicated that when sufficient liquid was available, more granule heating was strongly correlated with a higher degree of gelatinization, hence a better granule quality. The latter was also illustrated by SEM images: when comparing the image of run 11 (Fig. 4b, DoE 1) to run 7 (Fig. 6, DoE 2), it was seen that a higher degree of gelatinization was present with run 7. This can be explained by the intenser densification due to the modification of the screw configuration and the higher filling degree.

Table 3 Overview of the different run with fines fraction (%), oversized fraction (%), D50 (µm) and granule friability (%). (−): not possible to determine granule friability as insufficient granules > 250 µm were available. Run

Fines (%)

Oversized (%)

D50 (µm)

Friability (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

52.9 43.4 90.4 91.5 66.6 62.5 96.7 83.7 48.1 28.4 1.6 43.8 49.3 39.5 24.8 9.9 52.7 54.0 48.2 49.2

20.8 21.9 5.4 0.4 17.3 16.4 0.5 6.7 31.2 34.9 61.9 31.6 20.3 35.5 38.0 52.8 28.8 29.5 32.1 31.2

130.9 382.9 87.0 86.5 114.9 122.0 82.5 93.3 145.2 709.9 1136.8 237.0 158.1 352.1 581.4 936.0 142.9 139.7 184.4 170.8

– – – – – – – – 98.9 89.0 19.3 97.6 91.5 98.9 93.5 88.2 – – 97.7 96.4

proper quality. The evaluation of the granule properties and the microscopic images of the samples showed a correlation between the quality of the granules and the gelatinization degree of native pea starch: a better granule quality could be attributed to a higher degree of starch gelatinization. When starch was able to (partially) gelatinize in the barrel of the granulator, it enabled in situ binding properties (e.g. run 11). This required a low screw speed and high number of kneading elements which prolonged the residence time and densification of the powder. This positively affected the extent of compaction forces and friction experienced by the material in the barrel, enabling more granule heating. However, as microscopic analysis showed that complete starch gelatinization was not achieved under the evaluated settings, a second DoE was performed to assess if granule quality improved when complete starch gelatinization occurred during TSG. Moreover, as a high throughput (i.e. higher filling degree) positively affected the granule heating, a high throughput was used for all experiments of the second DoE. 3.1.2. Second screening design 3.1.2.1. Correlation between the in situ starch gelatinization and the granule quality. A Maltese cross was detected in runs 1 to 4 where the barrel temperature was set at low level which probably resulted in insufficient granule heating and consequently, no starch gelatinization took place. As it was previously shown that granule quality correlated

Fig. 4. SEM images for runs (a) 7, (b) 11, (c) 15 and (d) 16. The starch granules are indicated in red whereas the complete gelatinization of a starch granule is indicated in blue for run 11. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 6

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 5. Overview of the experimental runs of the second DoE with corresponding fines fraction (red bar), oversized fraction (green bar) and granule friability (dot). The presence of the Maltese cross for each run was indicated on the x-axis. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.1.3. Comparison between different types of native starches As complete gelatinization of pea starch was observed in the barrel of the granulator, the possibility of gelatinizing other native starches (maize, potato and wheat starch) during TSG was also evaluated. The gelatinization temperature for the different starches obtained with DSC, can be found in Table 4. Different gelatinization temperatures for the different starches are observed. This difference can be attributed to a combination of different factors: crystalline polymorphism, amylose content, chain length of amylopectin, lipid/protein content, molecular weight, etc. As these factors affect the energy characteristics of the different starches, each starch has its own unique gelatinization temperature (Bertoft, 2017; Bogracheva et al., 1998; Jane et al., 1999; Hizukuri et al., 1983; Sarko and Wu, 1978; Bule, 1998). It can be seen that the TSG gelatinization temperature corresponded with the range of gelatinization temperatures (onset – peak temperature) obtained with DSC (Table 4). Only the TSG gelatinization temperature of pea starch was lower compared to the gelatinization temperature obtained by DSC. Hizukuri et al. demonstrated that the X-ray diffraction patterns of starch granules with higher amylose contents showed poorer crystallinity. As pea starch contains the highest fraction of amylose of all tested starches, the higher degree of densification in

Table 4 Different starch types with corresponding DSC gelatinization temperature (°C), TSG gelatinization temperature (°C) and % of amylose. The TSG gelatinization temperature corresponded to the lowest barrel temperature resulting in granules with good quality. Type of starch

Gelatinization Temperature DSC (°C)

Maize starch Potato starch Pea starch Wheat starch

66.2 58.5 57.5 55.2

– – – –

71.1 64.1 64.2 61.7

TSG gelatinization Temperature (°C)

% amylose

70 60 55 55

24 20 40 25

the granulator barrel could explain the lower TSG gelatinization temperature (Hizukuri et al., 1983; AI PRESS). SEM images of the granulated samples were obtained for the different starch types and for both TSG temperatures. Fig. 7 illustrates SEM images of the samples granulated at an L/S-ratio of 0.23 for maize and potato starch. At a barrel temperature of 50 °C, the starch granule was still isolated but maize, potato and pea starch already showed some deformation. This indicated the start of the gelatinization process,

Fig. 6. (a) SEM and (b) PLOM-LP images obtained of granules from run 7 of the second DoE. 7

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 7. SEM images of maize and potato starch granules at a barrel temperature of 50 °C (left) and at the TSG gelatinization temperature (right). The starch granules are indicated in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

and no significant differences between the starch grades were observed. Furthermore, d50 shifted to higher values after TSG at higher temperatures, caused by the binding properties of the different starches after gelatinization. Taking the granule friability and the granule size into account, it can be concluded that also for these native starches (maize, potato and wheat starch) the gelatinization process took place and was highly correlated with the quality of the granules. This indicated the applicability of all starch types as in situ binder in continuous twin screw wet granulation.

resulting in low binding efficiency. Consequently, highly friable granules and lower values for d50 were obtained with maize, potato and pea starch. No significant differences between d50 values of pure DCP and d50 of DCP granulated with these starches were detected. In contrast, wheat starch granules were more deformed. The friability and d50 of wheat starch-based granules was affected by L/S-ratio as a lower granule friability and higher d50 were obtained at higher L/S-ratio. Possibly granule heating was insufficient at low L/S-ratio. As the available liquid contributed to an increased residence time of the wetted powder and therefore to granule heating, the presence of a higher amount may have resulted in gelatinization, yielding a lower granule friability and slightly higher d50. As wheat starch had the lowest onset gelatinization temperature (55.2 °C), it can be assumed that gelatinization already occurred at a barrel temperature of 50 °C and at high L/S-ratio, due to the higher granule heating caused by densification and friction forces. It was possible that due to the slightly higher onset gelatinization temperatures of pea, maize and potato starch, no effect of these starch types on granule properties was seen for each L/S-ratio when processed at 50 °C. Granule friability and values for d50 of all starch types are illustrated in Fig. 8. When the TSG gelatinization temperature was reached, all starch granules were completely gelatinized. This resulted in a high degree of interaction with the DCP particles. The higher barrel temperature, which was different for all starch types, caused sufficient granule heating to gelatinize the starch. The highest barrel temperature was required for the gelatinization of maize starch, which correlated with its high gelatinization temperature obtained from DSC. As gelatinization had occurred, the friability of all granules was of good quality (< 30%)

3.2. Starch as an in situ binder – Evaluation of a highly soluble formulation Gelatinization of native pea starch with DCP as excipient resulted in complete gelatinization and accordingly, in good granule quality. As DCP is practically insoluble in water, the granulation liquid was almost exclusively available for the gelatinization of starch. Therefore, it was of interest if gelatinization of pea starch could also occur in a formulation containing a highly soluble excipient, i.e. mannitol (Rowe and Sheskey, 2009). Experiments were performed at a barrel temperature of 55 °C, because previous results demonstrated gelatinization of pea starch with DCP at this temperature. The influence of a higher barrel temperature on the gelatinization process was also examined. The higher temperature could also be beneficial to enhance the water solubility of mannitol, which can increase the effective L/S-ratio (Ito and Kleinebudde, 2019). The high solubility and fast dissolution rate of mannitol defines that a lower L/S-ratio is required for efficient granulation (0.08 – 0.1425). 8

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

Fig. 8. Overview of the granule properties of the different starch types in function of L/S-ratio at (a) a barrel temperature of 50 °C and (b) at the TSG gelatinization temperature.

Fig. 9. Comparison of the granule friability at two different barrel temperatures in function of L/S-ratio (55 – 65 °C) for a pure mannitol and mannitol:pea starch (95:5) formulation.

an inadequate starch conversion. Therefore, no binder properties were established which was reflected in the higher granule friability with the mannitol/pea starch formulation. Furthermore, it is important to consider the solubility of the formulation in total. As the addition of a slightly higher water content might have resulted in starch gelatinization, native starches can be applicable binders with partially soluble excipients/active pharmaceutical ingredients.

Remarkably, at both process temperatures during TSG, a lower granule friability was observed using a pure mannitol formulation compared to mannitol/pea starch mixtures (Fig. 9). The differences in particle size distribution between both formulations were limited. SEM and PLOM images showed that complete starch gelatinization did not occur in the barrel of the granulator. Despite the inclusion of starch in the granulated particles, starch polarization (PLOM) and intact starch grain (SEM) were observed. As defined during the DCP study, sufficient liquid must be available for complete starch gelatinization. Since mannitol is processed at a lower L/S-range compared to DCP, less liquid was already available during TSG. Furthermore, because of the aqueous solubility of mannitol, many interactions between this excipient and the granulation liquid could be established. Accordingly, (part of) the available liquid interacted with the excipient, resulting in a lower availability granulation liquid to gelatinize the starch, hence in

4. Conclusion The study showed that, despite the short residence time, complete gelatinization of native pea starch in the barrel of the granulator was possible with DCP as model excipient. Furthermore, it was shown that the gelatinization process also took place with other native starches 9

International Journal of Pharmaceutics 571 (2019) 118760

L. Vandevivere, et al.

(maize, potato and wheat starch). The granule characterization highlighted the strong correlation between the degree of starch gelatinization and the granule quality. Higher barrel temperatures were needed in order to gelatinize starch, which might be beneficial for further processing steps (e.g. drying capacity). As experiments performed with a highly soluble excipient showed no gelatinization of starch, in situ starch gelatinization was only possible with low water soluble formulations. Additional research focusing on downstream processing and on more complex formulations could reveal the full potential of the starch binders. As high granule quality was obtained with a poorly soluble formulation, native starch is considered to be a very promising in situ binder for continuous twin screw wet granulation.

1016/j.powtec.2012.04.035. Hilbert E. Pea starch, a starch of high amylose content; 1945. Hizukuri, S., Kaneko, T., Takeda, Y., 1983. Measurement of the chain length of amylopectin and its relevance to the origin of crystalline polymorphism of starch granules. Biochim. et Biophys. Acta (BBA)-General Sub. 760, 188–191. Hoag SW. Pharmaceutical Granulation Processes, Mechanism and the Use of Binders; 2008; (July 2014). Ito, Akihiko, Kleinebudde, Peter, 2019. Influence of granulation temperature on particle size distribution of granules in twin-screw granulation (TSG). Pharm. Dev. Technol. 24 (7), 874–882. https://doi.org/10.1080/10837450.2019.1615089. Jane, J., Chen, Y.Y., Lee, L.F., Mcpherson, A.E., Wong, K.S., Radosavljevic, M., et al., 1999. Effects of amylopectin branch chain length and amylose content on the gelatinization and pasting properties of starch 1. Cereal Chem. 3258. Keleb, E.I., Vermeire, A., Vervaet, C., Remon, J.P., 2002. Continuous twin screw extrusion for the wet granulation of lactose. Int. J. Pharmac. 239, 69–80. Keleb, E.I., Vermeire, A., Vervaet, C., Remon, J.P., 2004. Twin screw granulation as a simple and efficient tool for continuous wet granulation. Int. J. Pharmac. 273, 183–194. Lund, D., Lorenz, K.J., 1984. CRC critical reviews in food science and nutrition influence of time, temperature, moisture, ingredients, and processing conditions on starch gelatinization. CRC Crit. Rev. Food Sci. Nutr. 20 (4), 249–273. Meier, Robin, Moll, Klaus-Peter, Krumme, Markus, Kleinebudde, Peter, 2017. Impact of fill-level in twin-screw granulation on critical quality attributes of granules and tablets. Eur. J. Pharm. Biopharm. 115, 102–112. https://doi.org/10.1016/j.ejpb.2017. 02.010. Rahman, B., Imam, M., Wahed, I., Khondkar, P., 2008. Effect of starch 1500 as a binder and disintegrant in lamivudine tablets prepared by high shear wet granulation. Pakistan J. Pharmac. Sci (May 2014). Remon, J.P., Visavarungroj, N., Herman, J., 1990. Crosslinked starch as binding agent wet granulation I. Conventional 59, 13–18. Rowe, R.C.R., Sheskey, P.J.S., 2009. Cook W. Sixth Edition, Handbook Pharmaceutical Excipients, pp. 1064. Saleh, M.F., Dhenge, R.M., Cartwright, J.J., Hounslow, M.J., Salman, A.D., 2015. Twin screw wet granulation : effect of process and formulation variables on powder caking during production. Int. J. Pharm. 496 (2), 571–582. https://doi.org/10.1016/j. ijpharm.2015.10.069. Sarko, B.A., Wu, H.H., 1978. The crystal structures of A-, B- and C-polymorphs of amylose and starch. Starch‐Stärke 3 (30). Stauffer, F., Ryckaert, A., Van Hauwermeiren, D., Funke, A., Djuric, D., Nopens, I., et al., 2019. Heat transfer evaluation during twin-screw wet granulation in view of detailed process understanding. AAPS PharmSciTech. Melkebeke B Van, Vervaet C, Remon JP. Validation of a continuous granulation process using a twin-screw extruder. 2008;356:pp. 224–30. Vanhoorne, V., Vanbillemont, B., Vercruysse, J., De, Leersnyder F., Gomes, P., 2016. Development of a controlled release formulation by continuous twin screw granulation : In fl uence of process and formulation parameters. Int. J. Pharm. 505 (1–2), 61–68. https://doi.org/10.1016/j.ijpharm.2016.03.058. Vercruysse, J., Córdoba Díaz, D., Peeters, E., Fonteyne, M., Delaet, U., Van Assche, I., De Beer, T., Remon, J.P., Vervaet, C., 2012. Continuous twin screw granulation: influence of process variables on granule and tablet quality. Eur. J. Pharm. Biopharm. 82 (1), 205–211. https://doi.org/10.1016/j.ejpb.2012.05.010. Vervaet, C., Remon, J.P., 2005. Continuous granulation in the pharmaceutical industry. Chem. Eng. Sci. 60, 3949–3957.

Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements The author would like to thank Mr Jean-Yves Pierquin and Mrs Isabelle Rambur for the help with the microscopic experiments. This work was supported by the INTERREG V 2 Mers Seas Zeeën Crossborder Cooperation Programme 2014–2020 (Project number 2S01-059 – project acronym IMODE). References AI PRESS. Starch Structure - Basic Carbohydrate Chemistry. In: AACC International. p. 1–11. Bertoft, E., 2017. Understanding starch structure: recent progress. Agronomy 7 (3), 56. http://www.mdpi.com/2073-4395/7/3/56. Bogracheva, T.Y., Morris, V.J., Ring, S.G., Hedley, C.L., 1998. The granular structure of Ctype pea starch and its role in gelatinization. Biopoly 45, 323–332. Bule, A., 1998. Starch. Gran.: Struct. Biosynth. 23, 85–112. Dhenge, R.M., Fyles, R.S., Cartwright, J.J., Doughty, D.G., Hounslow, M.J., Salman, A.D., 2010. Twin screw wet granulation : granule properties. Chem. Eng. J. 164 (2–3), 322–329. https://doi.org/10.1016/j.cej.2010.05.023. Djuric, D., Van, Melkebeke B, Kleinebudde, P., Remon, J.P., Vervaet, C., 2009. Comparison of two twin-screw extruders for continuous granulation. Europ. J. Pharmac. Biopharm. 71, 155–160. El, Hagrasy A.S., Hennenkamp, J.R., Burke, M.D., Cartwright, J.J., Litster, J.D., 2013. Twin screw wet granulation : In fl uence of formulation parameters on granule properties and growth behavior. Powder Technol. 238, 108–115. https://doi.org/10.

10