Evaluating Scale-Up Rules of a High-Shear Wet Granulation Process

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Evaluating Scale-Up Rules of a High-Shear Wet Granulation Process JING TAO, PREETANSHU PANDEY, DILBIR S. BINDRA, JULIA Z. GAO, AJIT S. NARANG Drug Product Science and Technology, Bristol-Myers Squibb, New Brunswick, New Jersey 08901 Received 3 February 2015; revised 24 April 2015; accepted 27 April 2015 Published online 25 May 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.24504 ABSTRACT: This work aimed to evaluate the commonly used scale-up rules for high-shear wet granulation process using a microcrystalline cellulose–lactose-based low drug loading formulation. Granule properties such as particle size, porosity, flow, and tabletability, and tablet dissolution were compared across scales using scale-up rules based on different impeller speed calculations or extended wet massing time. Constant tip speed rule was observed to produce slightly less granulated material at the larger scales. Longer wet massing time can be used to compensate for the lower shear experienced by the granules at the larger scales. Constant Froude number and constant empirical stress rules yielded granules that were more comparable across different scales in terms of compaction performance and tablet dissolution. Granule porosity was shown to correlate well with blend tabletability and tablet dissolution, indicating the importance of monitoring granule densification (porosity) during scale-up. It was shown that different routes can be chosen during scale-up to achieve comparable C 2015 granule growth and densification by altering one of the three parameters: water amount, impeller speed, and wet massing time.  Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci 104:2323–2333, 2015 Keywords: granulation; processing; porosity; compression; dissolution rate

INTRODUCTION Scale-up of high-shear wet granulation (HSWG) process is of high relevance to the pharmaceutical and food industries. The topic has been a subject of extensive research over the years and many publications exist to date on different ways to scale up this process.1–9 The various scale-up methodologies presented in literature can be broadly classified into two main categories: one that applies fixed mathematical rules to process parameters based on engineering principles from one scale to the other, and the other that empirically adjusts process parameters as needed to maintain the same granule attributes across scales. The main goal of establishing scale-up rules is to ensure that the particles experience similar granulation conditions across scales. It is believed that this similarity is best achieved by maintaining geometric, dynamic, and kinematic similarity across scales.9,10 Geometric similarity can be attained if the same equipment design (bowl height/diameter ratio, impeller design) is used across scales. Dynamic similarity refers to maintaining similar forces or collision energy experienced by the granules, and kinematic similarity refers to maintaining similar particle velocity inside the granulator. For a given equipment design, the force experienced by the particles and the particle velocities are mostly dictated by the impeller speed. Thus, significant effort has been put into establishing the best way to scale up impeller speed.5,10–13 The most commonly used scale-up rules related to impeller speed are the power law correlation governed by Eq. (1):  n T2 D1 = T1 D2

(1)

MATERIALS AND METHODS

where T is the impeller speed in revolutions per minute, D is the diameter of the impeller and subscripts 1 and 2 represent Correspondence to: Jing Tao (Telephone: +732-227-5025; Fax: +732-227-3986; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 104, 2323–2333 (2015)  C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association

different scales of granulators. A value of n = 1 in Eq. (1) corresponds to maintaining a constant impeller tip speed across scales, whereas a value of n = 0.5 corresponds to maintaining a constant Froude number across scales. In a high-shear granulator, the Froude number is essentially the ratio of angular acceleration to the gravitational acceleration. A value of n = 0.8 is an empirically derived exponential parameter that was shown to deliver similar shear stress across different scales, with successful scale-up results in some previous studies.5,14,15 It should be noted that most of the scale-up studies in literature attempt to achieve the best match of a limited set of granule properties. For example, some studies focus on granule particle size distribution (PSD) and dissolution, whereas others focus on granule PSD and flow. It has been shown in recent studies that PSD alone may not be a sensitive-enough response for describing a wet granulation process,16–19 and that other important granule properties such as pore size distribution and tabletability may vary without significant changes in PSD. Therefore, a study in which a comprehensive characterization of granules is conducted and all the responses are collectively considered is warranted. This forms the motivation of the current work. The aim of this work is to evaluate the commonly-used impeller speed and wet massing time related scale-up rules for a microcrystalline cellulose (MCC)–lactosebased formulation containing a water-insoluble drug. The evaluation included a comprehensive characterization of relevant granule properties (PSD, porosity, flow, tabletability) and the drug product quality attributes (tablet dissolution).

The formulation consisted of approximately 2:1 ratio of MCC and anhydrous lactose. To evaluate the impact of scale-up rules on drug dissolution, a poorly soluble drug, Compound-A (aqueous solubility: 0.040 mg/mL) was added to the formulation at 2.5% (w/w) level. The formulation composition is shown in Table 1.

Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

2323

2324

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Table 1. Formulation Composition Ingredient

Intragranular Microcrystalline cellulose Anhydrous lactose Hydroxypropylcellulose (Klucel EXF) Croscarmellose sodium Compound A Extragranular Croscarmellose sodium Magnesium stearate Total

Manufacturing Process % (w/w)

63.0 29.0 3.0 1.0 2.5 1.0 0.5 100.0

Microcrystalline cellulose and croscarmellose sodium were purchased from FMC Company (Philadelphia, Pennsylvania). Anhydrous lactose was purchased from Kerry Bio-Sciences (Norwich, New Jersey), hydroxypropyl cellulose from Ashland Specialty Ingredients (Wilmington, Delaware), and magnesium stearate from Mallinckrodt Chemicals (Saint Louis, Missouri). Compound-A was manufactured in house.

Experimental Design For this study, 1-L scale (200 g in batch size) was selected as the starting point of the scale-up to produce enough quantity of material for comprehensive granule characterization. At 10-L and 65-L scales, the batch sizes were selected to be 2 and 10 kg, respectively, so as to maintain equivalent fill volume across different scales. This study was conducted in two stages. At stage 1, three 1-L batches were manufactured with different water amounts (20%, 32%, and 40%, w/w, based on intragranular batch size) using process parameters selected based on the previous designof-experiment (DoE) study by Pandey et al.19 : 4.8 m/s impeller speed (628 rpm) and 30 sec wet massing time. The results were evaluated to determine the water amount to be used in the scale-up study. The selected set of parameters is a balanced choice between having sufficient robustness in the process and maintaining certain level of sensitivity to parameter changes. These 1-L batches also served as the small-scale reference batches when evaluating different scale-up rules. At Stage 2, 10-L and 65-L scale batches were manufactured by applying different scale-up rules. In this study, scale-up rules by impeller speed and by wet massing time were evaluated. For the impeller speed rules, while keeping the wet massing time constant at 30 s, the impeller speed was calculated by maintaining constant tip speed (m/s) (n = 1 in Eq. (1)), constant Froude number (n = 0.5), or constant empirical shear stress (n = 0.8). For the wet massing time rule, while keeping the impeller tip speed constant, the wet massing time was extended at the larger scale using the following calculation: t1 /t2 = D1 /D2 , where t is the wet massing time in seconds, D is the diameter of the impeller, and subscripts 1 and 2 represent different scales of granulators. At constant impeller tip speed, D1 /D2 = T2 /T1 . This leads to t1 /t2 = T2 /T1 or t1 ×T1 = t2 ×T2 , which is essentially keeping the total number of impeller rotation constant across scales. Tables 2 and 3 show the process parameters calculated using various rules. Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

The wet granulation experiments were conducted at three different granulator scales using 1-L Diosna high-shear granulator, 10-L PMA high-shear granulator, and 65-L PMA highshear granulator. The impeller speed during dry mixing was kept constant (5.0 m/s) for all batches. The chopper speed was set at 2000 rpm for 1-L Diosna granulator, and to “low” setting on the PMA granulators, which corresponded to approximately 1800 rpm. The impeller speed or wet massing time was varied based on the calculation using different scale-up rules. Water was used as the granulating liquid and was added to the bowl over a period of 3 min (constant for batches at all three scales) by dripping through a tube attached to a precalibrated pump. After wet massing, the material was passed through an 8-mesh screen. The granules were dried in a hot-air convection oven at 50°C to a target LOD (loss on drying) value of ±1% of the preblend LOD (2.5%–3.0%). The dried material was then riffled (Sieving riffler; Quantachrome Instruments, Boynton Beach, Florida) for an appropriate sample size for granule characterization. The remaining material was passed through a Quadro Comil (Quadro Engineering, Waterloo, Ontario) running at 2300 rpm using a 045R (1.1 mm opening) screen. The extragranular components were then mixed with the milled granules in a bin blender at approximately 50%–60% fill. This final blend was riffled to prepare samples for further characterization such as flow test and compaction analysis. The final blend was then compressed into tablets and submitted for dissolution and hardness tests. Characterization of Granulation and Tablets Particle Size Particle size distribution of the dried granules was measured by sieve analysis using an Allen Bradley Sonic Sifter (Allen Bradley, Milwaukee, Wisconsin) equipped with a set of six screens (# US 30, 40, 60, 80, 140, and 270) and a pan. Approximately 5 g sample was tested with a pulse setting of 5, sift setting of 5, and a total sifting time of 5 min. Porosity Pore volume distribution of the granulation was determined by Mercury Intrusion Porosimetry (MIP; AutoPore III; Micromeritics, Norcross, Georgia). A sample of 400 mg was used for analysis. Incremental and cumulative pore volumes were determined at different pressures ranging from 1 to 60,000 psi. The data were analyzed for pore sizes between the ranges of 0.1 and 10 :m. Flow R powder flow tester. About Flow was measured using Erweka 30 g sample was added to a 200-mL hopper at a stirring level of “2.” The outlet of the hopper was 11.3 mm in diameter. The result was taken as the average of three replicates.

Compaction R compaction Compaction data were generated using Stylcam simulator (Medel’Pharm, Beynost, France). A flat-faced round tooling (11.28 mm diameter) was used with direct cam setting. Three 400 mg tablets were compressed at five different solid fractions in the range of 0.75–0.95. The reported data are the average value of three replicates. For solid fraction calculations,

DOI 10.1002/jps.24504

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

2325

Table 2. Different Impeller Tip Speed Calculation at 10-L Scale (Parameters at 1-L Scale and 10-L Scale Were Subscripted with 1 and 2, Respectively). Wet massing time was kept constant at 30 s. Impeller Diameter (D1 , m) 0.146 0.146 0.146

Impeller Diameter (D2 , m)

Impeller Speed (rpm1)

Impeller Speed Scale-Up Rule

n Value in Eq. (1)

Impeller Speed (rpm2 )

Impeller Tip Speed at 10-L (m/s)

0.3 0.3 0.3

628 628 628

Constant tip speed Constant empirical shear stress Constant Froude number

1 0.8 0.5

305.6 353.0 438.1

4.80 5.54 6.88

Table 3. Wet Massing Time Scaled Up by the Ratio of Impeller Diameters at Different Scales. Impeller tip speed was kept constant at 4.8 m/s. Scale

1-L 10-L 65-L

Impeller Diameter (m) 0.146 0.3 0.5

Wet Massing Time (s) 30 62 103

tablet thickness was measured with Mitutoyo Digimatic thickness gauge. For dissolution measurements, the final blend of each batch was compressed into 100 mg weight tablets to a target hardness of 7 SCU using 15/64” round concave tooling. Dissolution The dissolution test was performed in 900 mL of 0.05 M sodium phosphate pH 6.8 with 0.05% SDS at 37°C, using USP Apparatus 2 (paddles) method at a rotation speed of 75 rpm. Samples were removed after 10, 20, 30, 45, and 60 min from test initiation and analyzed by HPLC at 280 nm. The reported data are the average ± standard deviation for n = 3 samples. The data were normalized to 100% dissolved at 60 min. Statistical Methods In a previous study, a full factorial DoE study was conducted at 1-L scale using three process parameters—impeller speed (4.0, 4.75, and 5.5 m/s), wet massing time (30, 60, and 90 s), and water amount (20%, 30%, and 40%, w/w). Each parameter in the DoE was varied at two levels—low and high—with four replicates at the center point. Additional experimental details of the 1-L DoE can be found in Pandey et al.19 In this work, the data from that study were used to build a statistical model using JMP 8.0 software to predict the porosity and extent of compaction at 1-L scale at the process conditions for which no experiment was conducted.

RESULTS Stage 1—Selection of Water Amount at 1-L Scale As described in the ‘Experimental Design’, Stage 1 study comprised of three batches manufactured at 1-L scale using different water amounts (20%, 32%, and 40%, w/w) while keeping impeller speed (4.8 m/s) and wet massing time (30 s) constant. The characterization results are presented as follows: in the order of the properties of the unmilled dried granules (particle size and granule porosity), the final blend (flow and compaction), and tablets (dissolution). DOI 10.1002/jps.24504

Figure 1. Particle size distribution (a) and porosity (b) of the unmilled granules of the 1-L batches granulated with different water amounts.

PSD of Unmilled Granules Figure 1a shows the PSD of the unmilled granules analyzed by sieve analysis. There was a trend of decreasing fines (particles <100 :m in size) and increasing large particles with increasing water amount. The geometric mean diameter calculated for the milled granules increased from 123 to 194 :m when water amount increased from 20% (w/w) to 40% (w/w). Porosity of Unmilled Granules Figure 1b shows the cumulative porosity profiles of the unmilled granules made with different water amounts at 1-L scale. Granule porosity decreased with increasing water amount within the range studied. The batch granulated with high water amount (40%, w/w) showed a more obvious decrease in granule porosity compared with the batches of low and medium water amounts (20% and 32%, w/w). Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

2326

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Flow of Final Blends

Dissolution

The flow of the final blend was acceptable (>7.0 g/min) for all the batches and appeared insensitive to changes of water amount. For this reason, flow will not be evaluated as a response in the following scale-up study.

Dissolution profiles were obtained for the tablets compressed to the same hardness from the final blends of the granulations manufactured with different water amounts. The drug dissolution became slower with increasing water amount during granulation (Fig. 2b). Water amount has been observed to be the most important factor for controlling granule growth in a HSWG process in the literature.20 Water activates the binder in the formulation by wetting the dry binder particles. On its own, water acts as a binder that facilitates coalescence upon particle collision and a lubricant that reduces the interparticulate friction. In a previous 1-L DoE study,19 we observed that the effect of impeller speed or wet massing time on the granule properties was a lot more pronounced at higher granulation water amount compared with medium or low water amount. In this study, the effect of water alone can be observed by comparing the 1-L scale batches, for which the water amount was varied while keeping impeller speed and wet massing time fixed. At increased water amount, the increasing particle size and decreasing porosity of the unmilled granules (Fig. 1) indicated the impact of water amount on the extent of granule growth and densification when the same shear (same impeller speed, granulation time, and wet massing time) was applied. Such impact was carried on through the downstream processes—the batch granulated with the highest water amount (40%, w/w) showed the lowest final blend tabletability and the slowest tablet dissolution (Fig. 2). Based on the results of the 1-L scale batches, 32% and 40% (w/w) water levels were selected as the target and high water levels to evaluate the scale-up rules. 32% (w/w) water was selected to represent the scale-up observations of an optimized and robust wet granulation process, whereas 40% (w/w) water was selected to represent a process with relatively high sensitivity to changes of the process parameters.

Compaction Performance of Final Blends Compaction performance was evaluated by tablet tensile strength versus compression force plot. Compaction was observed to be a sensitive response to changes of water amount. The batch of high water amount (40%, w/w) had the lowest tablet tensile strength at a given compression force (Fig. 2a), indicating some loss of granule tabletability at the high end of the water range. The compaction profile for the batch granulated with medium water amount (32%, w/w) was very comparable to that of the batch granulated with low water amount (20%, w/w), both of which were higher than that of the batch granulated with high water amount. In other words, the tabletability of the granules remained unchanged over the wide range of 20%–32% (w/w) water amount (representing a slightly undergranulated to optimally granulated situation), but decreased significantly between 32% and 40% (w/w).

Stage 2—Evaluation of Scale-Up Rules In Stage 2 experiments, the selected granulation process was scaled up to 10-L scale. For a given water amount, the characterization results were compared between different scale-up rules—constant tip speed rule (n = 1 in Eq. (1)), constant empirical stress rule (n = 0.8), constant Froude number rule (n = 0.5), or the wet massing time rule. The process parameters and characterization results of the 1-L and 10-L scale batches are listed in Tables 4 and 5 for 32% and 40% (w/w) water amounts, respectively. The characterization results (including granule particle size and porosity, final blend compaction performance, and tablet dissolution) are described in detail in this section. PSD of Unmilled Granules

Figure 2. Final blend tabletability profiles (a) and tablet dissolution profiles (b) of the 1-L batches granulated with different water amounts. Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

By Eq. (1), a low value of n corresponds to a high impeller speed at the larger scale. At 40% (w/w) water, the 10-L batches showed slightly smaller particle sizes than the 1-L batch. The amount of fines (particles of <100 :m in size) increased with increasing impeller speed or decreasing n value (Fig. 3b). The 10-L batch scaled up by maintaining constant Froude number (n = 0.5) showed the highest amount of fines (11%, w/w, more than the 1-L batch). The 10-L batch scaled by the wet massing time rule showed similar particle size as that scaled up by constant tip speed rule (n = 1). DOI 10.1002/jps.24504

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

2327

Table 4. Process Parameters and Characterization Results for Granulations Made with 32% (w/w) Water at 1-L and 10-L Scales Bowl Size

1-L

10-L

10-L

10-L

10-L

Impeller speed (m/s) Scale up power factor Wet massing time (s) Compaction—Final Blend Extent of compaction, 0.8–0.9 solid fraction (kPa) Dissolution (%) 10 min 30 min 45 min Granule Porosity Total intrusion volume (mL/g) Granule Particle Sizes Geometric mean diameter, (:m)

4.8 – 30

4.8 n=1 30

5.5 n = 0.8 30

6.8 n = 0.5 30

4.8 n=1 60

327

389

391

364

376

78 95 98

86 98 99

– – –

82 96 99

88 99 100

0.30

0.33

0.32

0.30

0.30

167

167

167

155

152

Table 5. Process Parameters and Characterization Results for Granulations Made with 40% (w/w) Water at 1-L and 10-L Scales Bowl Size

1-L

10-L

10-L

10-L

10-L

Impeller speed (m/s) Scale up power factor Wet massing time (s) Compaction—Final Blend Extent of compaction, 0.8–0.9 solid fraction (kPa) Dissolution (%) 10 min 30 min 45 min Granule Porosity Total intrusion volume (mL/g) Granule Particle Sizes Geometric mean diameter (:m)

4.8 – 30

4.8 n=1 30

5.5 n = 0.8 30

6.8 n = 0.5 30

4.8 n=1 60

269

402

392

332

329

66 90 99

79 95 98

– – –

74 94 97

81 96 99

0.27

0.34

0.30

0.29

0.27

194

183

184

168

176

Impeller speed has been known to have two-folded impact on granule growth during a HSWG process.16 A high impeller speed causes the particles to travel faster inside the granulator, promoting more particle–particle or particle–wall collisions to facilitate granule growth. On the contrary, the increased impeller speed may also promote the attrition among large particles and generate more fines in the granules. In this study, percentage fines increased when scaling up by constant Froude number, indicating that the attrition became the more dominant effect when scaling up by this rule. The differences between scale-up rules became even smaller for the batches granulated with 32% (w/w) water amount (Fig. 3a). The particle sizes for batches of 32% (w/w) water amount can be generally considered as comparable. Porosity of Unmilled Granules For batches granulated with 40% (w/w) water amount, the granule porosity increased when scaled up from 1-L to 10-L by constant impeller tip speed rule (n = 1) (Fig. 4b). When constant Froude number rule (n = 0.5, increased impeller speed at 10-L scale) was applied, the granule porosity of the 10-L scale batch decreased and became more comparable to that of the 1-L scale batch. The scale-up result by constant empirical shear rule (n = 0.8) was found to lie between the batches scaled up by n = 1 and n = 0.5 (Table 4). On the basis of the wet massing time rule, wet massing time was extended from 30 to 60 s at 10-L scale DOI 10.1002/jps.24504

(Table 3). The resulting granules showed comparable porosity to the 1-L scale batch and the 10-L scale batch of Froude number rule, suggesting that similar granule densification can be achieved at larger scale by increased impeller speed to maintain constant Froude number or longer wet massing time to maintain the same total number of impeller rotations. Much smaller changes in percentage porosity or total pore volume were observed for 32% (w/w) water amount batches for all scale-up rules tested (Fig. 4a). This is consistent with the robustness (comparable granule attributes in response to process parameter changes) expected at this water amount. In a separate experiment, a 10-L scale batch was manufactured with 32% (w/w) water using longer than usual wet massing time, during which granule samples were taken in small amounts (20 g) at different time points: 0, 30, 90, and 270 s. The samples were tray-dried and tested by MIP. Figure 5 shows that the granule porosity remained unchanged within the first 30 s of wet massing, the period that major granule growth has occurred as shown by the particle size data of the 10-L batches. As the wet massing continued, an obvious decrease in porosity was observed between 30 and 90 s, indicating some collapse of the pore structure of the granules during this time period. Additional wet massing to up to 3 min of total wet massing time did not cause further decrease in granule porosity, suggesting a plateau of granule densification during this period. This experiment demonstrated the effect of small changes in Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

2328

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Figure 3. Particle size distribution of the unmilled granules of the 10-L batches scaled up by different rules and the corresponding 1-L batch, granulated with 32% (w/w) water amount (a) and 40% (w/w) water amount (b), respectively.

Figure 4. Granule porosity of the 10-L batches scaled up by different rules and the corresponding 1-L batch, granulated with 32% (w/w) water amount (a) and 40% (w/w) water amount (b), respectively.

wet massing time on the microstructure of the granules and confirmed the hypothesis that desired granule porosity (and sizes) can be achieved, without changing impeller tip speed, by choosing the proper wet massing time for the process. Compaction Performance of Final Blends For the 40% (w/w) water batches, the tabletability (Fig. 6c) and compactability (Fig. 6d) of the final blends were plotted for the 1-L scale batch and the 10-L batches manufactured by applying different scale-up rules. Overall, the tabletability plots and compactability plots showed very consistent trends. Constant impeller tip speed rule generated granules of higher compaction performance at 10-L than at 1-L, in agreement with the higher granule porosity of that batch. The compaction profiles across different scales diverged more at higher compression forces. This is likely because the granules of low porosity reached a plateau of the highest achievable tensile strength compared to those of high porosity, for which a more porous structure allows for achieving higher tensile strength. When higher impeller tip speeds (n = 0.8 or 0.5) or longer wet massing time were used at 10-L scale, the tabletability of the granules decreased and approached that of the 1-L scale batch, following the same trend as observed with the granule porosity. For the batches granulated at 32% (w/w) water amount, the compaction performance of the 10-L scale batches was mostly Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

Figure 5. Granule porosity at different time points during extended wet massing of a 10-L batch granulated with 32% (w/w) water amount and 5.5 m/s impeller tip speed.

comparable between different scale-up rules (Figs. 6a and 6b) such that it is difficult to determine whether the small differences observed across scales were a much weaker manifestation of the same trend observed with 40% (w/w) water batches or were essentially the same among themselves. DOI 10.1002/jps.24504

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

2329

Figure 6. Final blend tabletability (a) and compactability (b) profiles of the 10-L batches scaled up by different rules and the corresponding 1-L batch, granulated with 32% (w/w) water amount. Final blend tabletability (c) and compactability (d) profiles of the 10-L batches scaled up by different rules and the corresponding 1-L batch, granulated with 40% (w/w) water amount.

Dissolution The dissolution results were considered not significantly different across scales given the inherent variability in dissolution measurements. The dissolution profiles of the 1-L scale batches were slightly lower compared with the 10-L scale batches for both water amounts (Fig. 7), consistent with the observed low granule porosity for the 1-L scale batches. Constant Froude number rule (n = 0.5) was shown to slowdown dissolution at 10-L scale and reduced the differences between 1-L and 10-L batches. Such differences were more obvious for 40% (w/w) water batches than 32% (w/w) water batches, similar to what has been observed with other granule properties such as PSD, tabletability, and porosity. Porosity–Tabletability–Dissolution Correlation It is noteworthy that when comparing the results of different scale-up rules, the differences in granule particle size between the batches became much less obvious after milling and final blending (data not shown). Nonetheless, the differences in granule porosity, an indication of the microstructures of the granules, continued to affect the critical performances of a batch in the consequent downstream processes: the batch of low-granule DOI 10.1002/jps.24504

porosity showed lower tabletability of the final blend and slower dissolution of the compressed tablets, and vice versa. The correlation between granule porosity and critical quality attributes (CQAs) such as tabletability and dissolution deserves more attention because granule porosity is typically not one of the standard characterizations for the granules. Thus, one should be very careful when using particle size alone as a response to assess granulation outcomes.18,19 It is interesting to note that the correlation between granule porosity, compaction, and dissolution observed at 10-L scale with the change of impeller speed or wet massing time when different scale-up rules were applied is similar to the correlation between granule porosity, compaction, and dissolution observed at 1-L scale, but with the change of water amount while keeping the impeller speed and wet massing time constant. Scale-Up to 65-L In order to verify the observations made at 10-L scale, two batches of 40% (w/w) water were manufactured at 65-L scale using the Froude number rule and the wet massing time rule. At 65-L scale, the wet massing time was extended to 100 s (Table 3). At the same impeller tip speed, similar tabletability was achieved with 100 s wet massing time at 65-L, compared Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

2330

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

to be lower than the 10-L batch also scaled up by constant Froude number rule, and closer to the 1-L scale batch. The different tabletability between the two batches at 65-L scale indicates the subtle difference between the Froude number rule and the wet massing time rule in matching the shear energy—the former by increasing the shear force (at increased impeller tip speed) over the same period of time and the latter by increasing the duration over which the same shear force is applied.

DISCUSSION

Figure 7. Tablet dissolution profiles of the 10-L batches scaled up by different rules and the corresponding 1-L batch, granulated with 32% (w/w) water amount (a) and 40% (w/w) water amount (b), respectively.

with 60 s wet massing time at 10-L (Fig. 8). By constant Froude number rule, the impeller tip speed was further increased at 65-L scale (to 8.9 m/s) while maintaining the wet massing time constant. The tabletability of the 65-L scale batch was observed

Figure 8. Final blend tabletability profiles of the batches granulated at 1-L, 10-L, and 65-L scales with 40% (w/w) water amount. Froude number rule and wet massing time rule were used to scale up from 1-L to 10-L and 65-L. Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

Scale-up of HSWG processes remains a persistent challenge in the pharmaceutical industry despite the extensive use of these processes in most companies. These challenges emanate from the short process time, usually on the order of 3–5 min, and relatively high-intensity process conditions (in terms of shear, particle velocities, and particle collision frequency and energy) compared with other batch processes such as low-shear wet granulation conducted in a planetary mixer. Historically, development of a high-intensity-short-duration process was designed to overcome the key challenge of densifying actives of low bulk density. The critical response variables for this goal were the manufacturability parameters of granule properties such as bulk density, PSD, flow, and sticking tendency. In this work, scale-up of a HSWG process was studied at 1-L, 10-L, and 65-L scales. These scales represent the common batch sizes utilized in HSWG during early and pilot scale product development. The cost and availability of API and the impact of batch loss (on project timelines and deliverables such as clinical supplies or commercial launch) increases as one scales up the process to larger scales for late-stage clinical and commercial drug product manufacture. Thus, scale-up of the HSWG processes is a high-risk operation where the number of batches one can manufacture at larger scales is increasingly limited by material constraints. At the same time, regulatory quality-bydesign guidance mandate generating a scientifically sound understanding of the process, such as by controlled DoE studies,21 at relevant scales to demonstrate robustness in the process. In this study, we observed that granule porosity and compaction performance were higher (indicating lower granule densification) at 10-L scale than at 1-L scale, if the process parameters are kept constant during scale-up (Figs. 4 and 6). This is consistent with general observations of scale-up from the pilot scale to the commercial scale. Thus, the understanding of the mathematical scale-up rules generated at these (lab and pilot) scales was deemed representative of the scale-up path of the process from the lab to the commercial scale. It is increasingly being realized that granule porosity is the overriding mechanism22 impacting drug product CQAs such as drug release or dissolution23 and bioavailability.24 In fact, porosity has been recognized recently as the mechanistic basis of the effect of process parameters on drug product quality attributes in HSWG.20 Therefore, PSD alone may not be an adequate response parameter to assess granulation outcomes.18,19 In this study, a comprehensive evaluation of granulation characteristics was undertaken to include granule densification (porosity) in addition to PSD when evaluating different scale-up rules. Correlation between granule porosity, compaction, and dissolution was observed at both 10-L scale and 1-L scale. The change of granule porosity and tabletability of the final blends were compared in Figure 9 with the respective changes DOI 10.1002/jps.24504

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

2331

Figure 9. The change of granule porosity (a–c) and extent of compaction (d–f) with the respective change of impeller speed, wet massing time, and water amount at 1-L and 10-L scale while keeping the other two parameters constant (values are specified in each figure). The blue symbols represent the 10-L data from the experiments. The red symbols represent 1-L data from the experiments (open diamonds) or from the JMP statistical model (filled diamonds). The green arrows in (a) and (b) represent two different routes for scale-up: by the vertical arrows, all process parameters are kept constant across scales, leading to higher granule porosity and extent of compaction at the larger scale; by the horizontal arrows, tip speed (a) or wet massing time (b) are increased at the larger scale to achieve comparable granule porosity and compaction performance across scales. Table 6. Statistical Model of Granule Porosity and Extent of Compaction of Final Blend as a Function of Impeller Tip Speed, Water Amount, and/or Wet Massing Time Parameter Estimates Term Granule Porosity Intercept Tip speed (4, 5.5) Water amount (20, 40) Wet massing time (30, 90) Tip speed,* water amount Extent of Compaction Intercept Tip speed (4, 5.5) Water amount (20, 40) Wet massing time (30, 90) Tip speed,* water amount Tip speed,* wet massing time Water amount,* wet massing time

Estimate

Standard Error

t Ratio

Prob > |t|

0.2636667 −0.012375 −0.043125 −0.014875 −0.014875

0.006902 0.008453 0.008453 0.008453 0.008453

38.20 −1.46 −5.10 −1.76 −1.76

<0.0001* 0.1866 0.0014* 0.1218 0.1218

3.211436 3.933189 3.933189 3.933189 3.933189 3.933189 3.933189

78.73 −8.04 −8.78 −3.32 −6.54 −2.95 −6.42

<0.0001* 0.0005* 0.0003* 0.0210* 0.0013* 0.0320* 0.0014*

252.83333 −31.6375 −34.5375 −13.0625 −25.7125 −11.5875 −25.2375

of impeller speed, wet massing time, and water amount at both 1-L and 10-L scales. The data of the 10-L scale were obtained directly from the Stage 2 experiments (represented by blue dots). Where available, the data of the 1-L scale were from the experiments conducted in Stage 1 and the 1-L DoE in a previous study19 (represented by open diamond symbols). For process conditions at which no experiment was conducted, a statistical model (see Materials and Methods for details) was used to predict the granule properties (Table 6) (represented by filled diamond symbols). This enabled a direct comparison between 1-L and 10-L scales at the same process conditions. DOI 10.1002/jps.24504

In Figure 9, all the process parameters impact porosity (granule densification) in the same direction, shown by the negative slopes of the trend lines connecting the data of the same scale in Figures 9a–9c. Thus, increasing any of the three process parameters—impeller tip speed, wet massing time, or water amount (w/w %)—can lead to comparable impact on granulation. This correlated well with the effect of process parameters on the tabletability (negative slopes in Figs. 9d–9f). These consistent trends indicated the ubiquity of the porosity–compaction–dissolution correlation discussed in the Results section, in that the densification of the granules (reduction in porosity) has a strong impact on its Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

2332

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

downstream performance regardless of how that densification is achieved. These results further underscore the importance of monitoring and controlling granule porosity as an in-process material quality attribute. Additionally, in situations when one or two of these parameters are limited to a certain value(s) by other restrictions (e.g., equipment impeller speed limitations), the rest of the parameters can still be carefully selected to achieve similar granulation outcome during scale-up. Figure 9 also showed that both granule porosity and final blend tabletability were higher at 10-L scale than at 1-L scale for any given set of the process parameters. This suggests that the particles are less granulated if all the process parameters are maintained constant during scale-up. Such a scale-up route can be schematically represented in each figure by a vertical arrow from the 1-L data point to the 10-L data point (shown in Figs. 9a and 9b as two examples). It has been shown in the literature that constant impeller tip speed can maintain kinematic similarity at different scales by providing comparable magnitudes of the internal particle shear flow, but cannot maintain dynamic similarity because the cumulative particle collision energy per unit time significantly decreases at increased scales.9 In a recent study by Nakamura et al.,9 it was shown through discrete element modeling simulations that dynamic similarity across scales can be achieved by increasing the wet massing time upon scale-up, to match the cumulative particle collision energy per unit time across scales. Wet massing time is a process parameter of significant importance that has generally not been given due consideration during scale-up studies reported in the literature.20 Recent work has shown that the kinetics of granule densification are most rapid during the wet massing time, presumably because of the availability of all the fluid in the system. Rekhi et al.10 recommended to maintain the same impeller tip speed (in m/s) across scales and scale wet massing time based on the inverse ratio of the impeller speeds in rpm in order to keep the same total number of impeller rotations across scales. This is essentially equivalent to scaling wet massing time by the ratio of the impeller diameters at constant tip speed. In this study, using this scale-up methodology, the wet massing time was increased from 30 s at 1-L scale to 60 s at 10-L scale while maintaining impeller tip speed and water amount constant, through which granules of comparable porosity were achieved at the larger scale (a scale-up route schematically shown in Fig. 9b by the horizontal arrow). A more commonly used approach to scale up is by the tip speed-related rules (Table 2). The literature has several examples in which scaling-up by constant tip speed worked better than the constant Froude number rule, along with examples showing the opposite trend,5–7,9–13,25,26 and cases where the constant shear-rule led to successful scale-up.5,14,15 In this study, we evaluated different values of the exponential parameter in Eq. (1), including Froude number rule (n = 0.5), constant tip speed rule (n = 1), or empirically understood constant shear rule (n = 0.8). We observed that, while keeping all other parameters constant, constant impeller tip speed produced granules of slightly smaller particle size and higher porosity at 10-L scale, suggesting a less extent of granule growth and densification compared with 1-L. Although increased compaction performance or high-dissolution rate usually associated with less granulated particles may not be undesirable during scale-up, there are occasions when it is important to have good reproTao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015

ducibility during scale-up. When higher impeller speed was used at the large scale by applying constant empirical stress rule or constant Froude number rule, the granule properties became more comparable across scales. This scale-up route is schematically shown by the horizontal arrow in Figure 9a. The similarity between Figures 9a and 9b indicates that a successful scale-up requires that a good match of the shear stress experienced by the particles at the small scale is achieved by using higher impeller speed (such as Froude number rule) or by using the same impeller speed, but longer mixing time (wet massing rule) at the large scale. Alternatively, these two rules can be used to reproduce, at 1-L scale, the granulation observations made at large scales for process development or troubleshooting purposes. When used in scale-down calculations, these rules essentially demands for lower impeller speed or shorter wet massing time at 1-L scale so as to compensate for the more aggressive densification experienced by the granules in a 1-L mixer. One of the practical challenges with using Froude number rule in scale-up or scale-down is that, depending on the magnitude of the differences between the scales, this rule may require impeller speed values that are not achievable in some granulators. When such limitations exist, wet massing time rule can be a good alternative route for scale-up.

CONCLUSIONS This study provided a holistic investigation of the effect of granulation process parameters on drug product CQA-relevant inprocess granule attributes. Different scale-up rules were evaluated using a low drug loading formulation at lab and pilot scales of manufacture. Our findings suggest that process parameters adjusted to (1) constant Froude number at fixed wet massing time and (2) the extended wet massing time at constant impeller tip speed (n = 1) produce comparable granules across different scales. Both approaches increased the amount of energy applied to the powders at the higher scale by either mixing them at a faster tip speed than at 1-L scale or mixing them at the same tip speed as 1-L, but for a longer duration. In addition, granule porosity was found to correlate well with final blend compaction and tablet dissolution. This further underscored the importance of monitoring and controlling granule porosity, in addition to the traditionally well-recognized granule attributes such as PSD, flow, density, and sticking.

REFERENCES 1. Aikawa S, Fujita N, Myojo H, Hayashi T, Tanino T. 2008. Scaleup studies on high shear wet granulation process from mini-scale to commercial scale. Chem Pharm Bull 56(10):1431–1435. 2. Ameye D, Keleb E, Vervaet C, Remon JP, Adams E, Massart DL. 2002. Scaling-up of a lactose wet granulation process in Mi-Pro high shear mixers. Eur J Pharm Sci 17(4–5):247–251. 3. Faure A, Grimsey IM, Rowe RC, York P, Cliff MJ. 1999. Applicability of a scale-up methodology for wet granulation processes in Collette Gral high shear mixer-granulators. Eur J Pharm Sci 8(2):85–93. 4. Faure A, York P, Rowe RC. 2001. Process control and scale-up of pharmaceutical wet granulation processes: A review. Eur J Pharm Biopharm 52(3):269–277. 5. Horsthuis GJB, van Laarhoven JAH, van Rooij RCBM, Vromans H. 1993. Studies on upscaling parameters of the Gral high shear granulation process. Int J Pharm 92(1–3):143–150. DOI 10.1002/jps.24504

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

6. Iskandarani B, Shiromani PK, Clair JH. 2001. Scale-up feasibility in high-shear mixers: Determination through statistical procedures. Drug Dev Ind Pharm 27(7):651–657. 7. Landin M, York P, Cliff MJ, Rowe RC. 1999. Scaleup of a pharmaceutical granulation in planetary mixers. Pharm Dev Technol 4(2):145– 150. 8. Litster JD. 2003. Scaleup of wet granulation processes: Science not art. Powder Technol 130(1–3):35–40. 9. Nakamura H, Fujii H, Watano S. 2013. Scale-up of high shear mixer-granulator based on discrete element analysis. Powder Technol 236(0):149–156. 10. Rekhi G, Caricofe R, Parikh D, Augsburger L. 1996. A new approach to scale-up of a high-shear granulation process. Pharm Technol 20(10):1–10. 11. Hassanpour A, Kwan C, Ng B, Rahmanian N, Ding Y, Antony S, Jia X, Ghadiri M. 2009. Effect of granulation scale-up on the strength of granules. Powder Technol 189(2):304–312. 12. Holm P. 1987. Effect of impeller and chopper design on granulation in a high speed mixer. Drug Dev Ind Pharm 13(9–11):1675–1701. 13. Rahmanian N, Ng B, Hassanpour A, Ghadiri M, Ding Y, Jia X, Antony J. 2008. Scale-up of high shear mixer granulators. KONA 26:190–204. 14. Tardos GI, Hapgood KP, Ipadeola OO, Michaels JN. 2004. Stress measurements in high-shear granulators using calibrated “test” particles: Application to scale-up. Powder Technol 140(3):217–227. 15. Tardos GI, Khan MI, Mort PR. 1997. Critical parameters and limiting conditions in binder granulation of fine powders. Powder Technol 94(3):245–258. 16. Badawy SI, Menning MM, Gorko MA, Gilbert DL. 2000. Effect of process parameters on compressibility of granulation manufactured in a high-shear mixer. Int J Pharm 198(1):51–61. 17. Chaudhury A, Wu H, Khan M, Ramachandran R. 2014. A mechanistic population balance model for granulation processes: Effect of process and formulation parameters. Chem Eng Sci 107(0):76–92.

DOI 10.1002/jps.24504

2333

18. Pandey P, Sinko PD, Bindra DS, Hamey R, Gour S, Vema-Varapu C. 2013. Processing challenges with solid dosage formulations containing vitamin E TPGS. Pharm Dev Technol 18(1):296–304. 19. Pandey P, Tao J, Chaudhury A, Ramachandran R, Gao JZ, Bindra DS. 2013. A combined experimental and modeling approach to study the effects of high-shear wet granulation process parameters on granule characteristics. Pharm Dev Technol 18(1):210–224. 20. Badawy SI, Narang AS, LaMarche K, Subramanian G, Varia SA. 2012. Mechanistic basis for the effects of process parameters on quality attributes in high shear wet granulation. Int J Pharm 439(1–2):324– 333. 21. Woyna-Orlewicz K, Jachowicz R. 2011. Analysis of wet granulation process with Plackett-Burman design—Case study. Acta Pol Pharm 68(5):725–733. 22. Narang AS, LaMarche K, Subramanian G, Lin J, Varia S, Badawy S. 2011. Granule porosity is the overriding mechanism controlling dissolution rate of a model insoluble drug from an immediaterelease tablet. AAPS Annual Meeting,http://abstracts.aaps.org/ SecureView/AAPSJournal/radg1axf1sg.pdf, last accessed on 5/13/2015. 23. Wolfe S, Pafiakis S, Tang D, Jennigs S, Abebe A, Gao Z, Varia S, Narang AS. 2013. Effect of tablet pore size distribution on dissolution upon accelerated storage conditions. AAPS Annual Meeting,http:// abstracts.aaps.org/SecureView/AAPSJournal/radfxj1ebxj.pdf, last accessed on 5/13/2015. 24. Pandey P, Hamey R, Bindra DS, Huang Z, Mathias N, Eley T, Crison J, Yan B, Perrone R, Vemavarapu C. 2014. From bench to humans: Formulation development of a poorly water soluble drug to mitigate food effect. AAPS PharmSciTech 15(2):407–416. 25. Knight PC, Seville JPK, Wellm AB, Instone T. 2001. Prediction of impeller torque in high shear powder mixers. Chem Eng Sci 56(15):4457–4471. 26. Rahmanian N, Ghadiri M, Ding Y. 2008. Effect of scale of operation on granule strength in high shear granulators. Chem Eng Sci 63(4):915–923.

Tao et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:2323–2333, 2015