G Model

IJP 14297 1–11 International Journal of Pharmaceutics xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

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

1 2

Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework

3 Q1

Elena Stocker a , Gregor Toschkoff a , Stephan Sacher a , Johannes G. Khinast a,b, *

4 5

a b

Research Center Pharmaceutical Engineering GmbH, Inffeldgasse 13, Graz 8010, Austria Institute for Process and Particle Engineering, Graz University of Technology, Inffeldgasse 13, Graz 8010, Austria

A R T I C L E I N F O

A B S T R A C T

Article history: Received 18 July 2014 Received in revised form 21 August 2014 Accepted 27 August 2014 Available online xxx

The purpose of this study is to evaluate the use of computer simulations for generating quantitative knowledge as a basis for risk ranking and mechanistic process understanding, as required by ICH Q9 on quality risk management systems. In this specific publication, the main focus is the demonstration of a risk assessment workflow, including a computer simulation for the generation of mechanistic understanding of active tablet coating in a pan coater. Process parameter screening studies are statistically planned under consideration of impacts on a potentially critical quality attribute, i.e., coating mass uniformity. Based on computer simulation data the process failure mode and effects analysis of the risk factors is performed. This results in a quantitative criticality assessment of process parameters and the risk priority evaluation of failure modes. The factor for a quantitative reassessment of the criticality and risk priority is the coefficient of variation, which represents the coating mass uniformity. The major conclusion drawn from this work is a successful demonstration of the integration of computer simulation in the risk management workflow leading to an objective and quantitative risk assessment. ã 2014 Published by Elsevier B.V.

Keywords: Quality by design Process understanding ICH Q9 ?quality risk management Computational simulation Discrete element method Tablet coating

6

1. Introduction

7

Quality by design (QbD) has by now become a well-known and widely accepted science- and risk-based regulatory approach that focuses on safety, efficacy and quality throughout the product’s life cycle. It is increasingly used in the pharmaceutical industry to implement the vision of cost-efficient, market-oriented and high-quality products for patients (Aksu et al., 2012; European Medicines Agency, 2009; Kessel, 2011). Within the QbD framework risk-based approaches for quality assurance play an important role. Applying the risk-based approach via quality risk management (QRM) leads to a continuous improvement of process performance and product quality (International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2008). Furthermore an effective application of QRM throughout the product’s life cycle offers increased flexibility and reduced risk of product failure, as described in the International Conference on Harmonization (ICH) Q9 document (World Health Organization, 2010). Since this document was published, the

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

* Corresponding author at: Graz University of Technology, Institute for Process and Particle Engineering, Inffeldgasse 13, Graz 8010, Austria. Tel.: +43 316 87330400; fax: +43 316 8731030400. E-mail address: [email protected] (J.G. Khinast).

regulatory requirements for assessing, controlling, reviewing, and communicating risks have become an integral part of the pharmaceutical quality system (International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005), supported by the implementation of decision loops to facilitate risk control (International Organization for Standardization, 2007). Clearly, the effectiveness of the qualitative specification and the quantitative estimation of risks strongly depend on the underlying data and experience of individuals involved in the process. According to this fact, QRM initially identifies risks linked to the potentially critical quality attributes (CQAs), which characterize the products’ quality (International Organization for Standardization, 2009a). These potentially CQAs (e.g., content uniformity of dosage units) are defined as “quantifiable and potentially critical characteristics” if a negative influence on the intended product’s efficacy, quality and patient safety may occur (European Medicines Agency, 2009). Understanding of the CQAs and the associated risks of failure is thus a critical part of the QbD approach. However, tools available for the risk assessment process are, to a large extent, based on qualitative decisions involving experts in the field (e.g., via FMEA or related tools), and a quantitatively exact framework is missing (Miláet al., 2012; Zimmermann and Hentschel, 2011). Therefore, in this publication we attempt to present a more quantitative

http://dx.doi.org/10.1016/j.ijpharm.2014.08.055 0378-5173/ ã 2014 Published by Elsevier B.V.

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

G Model

IJP 14297 1–11 2 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100

E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

approach, based on mechanistic modeling and simulations. As an example process, the coating of tablets was chosen. Specifically, an active coating process (i.e., the coating contains an API) was considered. Among other dosage forms, the ICH Q6A guideline describes the quality attributes (QAs) of solid oral dosage forms, such as tablets with active coating. One QA is the uniformity of dosage units, which can either be represented by the mass of the dosage form or the content of the drug substance in the dosage form (International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 1999). In our work, the focus is the potentially CQA “content uniformity of drug substance in the active coating”, which is represented by the coating mass uniformity or inter-tablet coating variability. Typically, experimental methods (mostly involving lab-scale equipment) have been used to study process influences on product quality and QAs. However, such studies are time-consuming and hard to perform on production-scale equipment (coating studies in full-scale coaters are highly demanding). In contrast, novel modeling and simulation approaches make it possible to investigate the impact of process conditions on product quality based on first principles and (which is a major advantage) at any desirable scale with any equipment type (Suzzi et al., 2012, 2010; Toschkoff and Khinast, 2013; Toschkoff et al., 2012). Thus, in a quantitative risk assessment process, mechanistic simulations can play an important role: Such an “improved” risk assessment process would involve first the identification of failure modes based on an initial risk assessment. Second, a quantitative criticality assessment through risk priority evaluation would occur, both of which can be based on simulations. Lastly, product and process characterization could (at least partially) be guided by simulations. Thus, the gained mechanistic understanding may also support the process parameter (PP) definition and the screening studies within the QRM framework. This is a new approach as demonstrated in Fig. 1. Although the ICH Q11 document already mentions the application of modeling and simulation for QRM during the development of manufacturing processes, model-based simulations have so far not been used to support the QRM process in a quantitative manner (European Medicines Agency, 2011; International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2004). In this work we intend to support a quantitative risk assessment process by ensuring objectivity through the application of highfidelity computer simulations (Fig. 1). The stepwise approach integrates computer simulation in DoE-based process parameter screening studies and in the verification of quantitative risk assessment. This stepwise approach was demonstrated by analyzing a tablet active-film coating process performed in a pan coater representing a common unit operation in the pharmaceutical industry. 2. Material and methods

101

2.1. The coating process

102

Most commonly, coated tablets are produced via drum coating, during which tablet cores are placed in a continuously rotating drum and sprayed from above. Due to the rotation, the tablets periodically enter and leave the spray zones of each spray nozzle. Concurrently, the coating layer is dried by evaporating the solvent with heated air (Cole et al., 1995). Perfect inter-tablet coating variability is achieved, if every tablet enters the spray zone for the same amount of time (Chen et al., 2010; Dubey et al., 2011; Toschkoff et al., 2012). Since this is not the case in a realistic process, perfect coating uniformity is never achieved. A

103 104 105 106 107 108 109 110 111

minimization of the inter-tablet coating variability is, however, an important goal of process development nowadays (Tobiska and Kleinebudde, 2003). Due to this fact, experiment- and simulation-based investigations focused on the coating and tablet mixing operations in various coating devices. For example, an experimental study by Porter et al. (1997) studied the coating uniformity, percent loss on drying and coating process efficiency to optimize the drug release profile based on DoE principles, including drying inlet air temperature, fluid spray rate, atomizing air pressure, pan speed and number of spray guns. In the work of Kalbag and Wassgren (2009) and Kalbag et al. (2008) the impact of tablet residence time within the spray zone on inter-tablet coating variability was studied as a function of the PPs pan speed, fill level and coating time. Furthermore, a computational study by Dubey et al. (2011) focused on the design of spray pattern amongst others. Toschkoff et al. (2012) identified the inlet air temperature, air flow rate, pan speed, spray nozzle position and spray nozzle direction as PPs with significant impact on spray loss, using numerical computational fluid dynamics (CFD) modeling. The sensitivity of the relative standard deviation (RSD) of the content uniformity to the variation of various critical coating PPs was evaluated by Chen et al. (2010). Just et al. (2013a) evaluated the PPs fill level, pan speed, spray rate, spray time and spray pressure for lab and pilot scale. One of the important conclusions was that the increase of the spray nozzle number in the pan coater leads to a significant impact on coating uniformity (Just et al., 2013a). Among others, these scientific findings provide the basis for the cause and effect analysis in this study.

112

2.2. Identification of process parameters

141

First, an initial identification of direct or indirect causes for non-uniformity of the tablet coating mass (Fig. 2) was performed. PPs for tablet mixing and coating, such as air flow, spray pattern, pan speed and fill level, as well as equipment design parameters, such as the spray nozzle position, angle and distance to the tablet bed were considered (Fig. 2). This provides the basis for the assessment of potential failure modes via a quantitative ranking of the severity, the likelihood of occurrence and their detectability, described in the next section.

142

2.3. Initial risk assessment via PFMEA

152

2.3.1. Process failure mode and effects analysis (PFMEA) Since its introduction as a risk assessment tool into pharmaceutical quality systems, many different approaches and standards have been published (International Electrotechnical Commission, 2006; International Organization for Standardization, 2009b; Liu et al., 2013; Stamatis, 2003). In general, a PFMEA uses the risk priority number (RPN) to classify the criticality of failure modes and the need of corrective actions. It is calculated by multiplying the assigned values for severity (S), occurrence (O) and detectability (D). For each class, scores are given by a panel of experts. Severity represents the effect of a failure that occurs during the process on the product quality or patient safety. It is classified as a numerical value according to the impact of the negative consequences. The occurrence characterizes the frequency of appearance of a failure. The detectability shows if the failure is easy to identify (Stamatis, 2003). For each of the three risk factors, the levels and their requirements are defined in a risk assessment catalog (Table 1). In this study we assigned odd numbers from 1 to 9, with five levels for each risk factor (International Organization for Standardization, 2009a).

153

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140

143 144 145 146 147 148 149 150 151

154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172

G Model

IJP 14297 1–11 E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx 173 174 175 176 177

2.4. Quantitative criticality assessment for PPs through risk priority evaluation In order to classify the failure mode criticality, an RPN is defined, as mentioned above. The higher the RPN, the more critical is the failure mode. However, also a critical level has to be defined.

3

A commonly used strategy is to define this critical level as the product of the three average category levels i.e., 5  5  5 = 125. If a failure mode has a critical RPN (125), the implementation of corrective actions for the reduction to an acceptable ranking is mandatory. This leads to a re-evaluation of the ranking and the recalculation of the RPN.

Fig. 1. Overview of the stepwise quantitative risk assessment approach based on an integration of computer simulation in the quality risk management workflow.

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

178 179 180 181 182 183

G Model

IJP 14297 1–11 4

E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

Fig. 2. Cause and effect analysis – identification of critical PPs which directly or indirectly influence coating mass uniformity. Process-related parameters qualify tablet mixing and coating operations. Equipment design-related parameters characterize the spray nozzle.

184 185 186 187 188 189 190 191 192 193 194

In addition, the PPs are divided into two types, which describe the criticality. In this study the PFMEA table contains a column “PP type”, in which the PP is either categorized as general (GPP) or critical (CPP) (Table 2). This means that the PP type specifies the direct or indirect impact on the QA or process performance. Therefore, it is generally assumed that any criticality assessment must be based on prior knowledge or experimental data. For the analysis performed here, a tabular PFMEA was used and an initial assessment of the severity, occurrence and detectability for potential CPPs of tablet mixing and coating was generated. Furthermore, based on the risk priority number (RPN) the type of

each PP was defined. For each defined PP, the potential failure modes and potential effects of failure are described and the severity is assessed based on the assessment catalog illustrated in Table 1. Another column of the PFMEA illustrated in Table 6 discusses occurrence, based on the defined potential causes or mechanisms of failure. The third risk factor, detectability, is classified based on the design to control prevention and detection. If the parameter is classified as potential CPP based on the mentioned risk factors, recommended actions are mandatory. In some cases a lack of prior knowledge about a GPP requires fundamental studies to ensure that the criticality assessment is

Table 1 The risk assessment catalog used for the process failure mode and effects analysis: the ranking is based on a five-level scale (odd numbers from one to nine) for each risk factor (severity, occurrence and detectability of failure). Furthermore each ranking level is described and defined in detail. Ranking

Description

Severity (S) 1 Low 3 5 7 9

Low to moderate Moderate Moderate to high High

Occurrence 1 3 5 7 9

(O) Rare/uncommon Low Moderate High Very high

Detectability of failure (D) 1 Failure will always be detected (offline, online, inline measurement) 3 Detection highly likely (offline measurement) 5 Occasionally not detected – failure may be missed sometimes Probably not detected – failure may be 7 missed often Impossible to detect (no equipment/ 9 method)

Definition CQA not affected, severity of potential failure mode indistinguishable, failure is not noticed, no impact on product quality. Low impact on CQA failure can easily be prevented, no time loss. Failure mode causing moderately severe impact in CQA. Moderate to high impact on CQA and therefore on product quality. Failure mode has high impact on CQA and therefore extremely severe impact on product quality and patient safety, patient hazard (microbiological infections, hypersensitivity, immune response).

Never occurs. Very low probability of occurrence. Occurs every now and then. Moderate probability. Very common.

Permanent technical monitoring and alarm system exists; monitored by at least one or more sensors; high probability of detection of failure. Error is obvious, regular but not permanent supervision, several mutually independent ways to discover the failure, moderate probability of detection, visual check by operator. Just a few independent ways to discover the error, low probability of detection of failure. Representative sampling, very low detectability of failure. Failure cannot be checked because of technical reasons, random testing; failures can be detected only by qualified personnel, possibility of discovering unlikely.

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

195 196 197 198 199 200 201 202 203 204 205

G Model

IJP 14297 1–11 E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

5

Table 2 Process parameter type categories (abbreviation, name, description of category) (CMC Biotech Working Group, 2009).

206 207 208

Abbreviation

Name

Description of category

CPP

Critical process parameter

GPP

General process parameter

CPP must be investigated within process characterization studies and will be included in the design space. This parameter impacts a QA and has therefore a direct impact on product quality. CPP must be controlled tightly and is characterized by limited robustness. GPP is an adjustable parameter that has no significant effect on product quality or process performance. The classification must be proven within process characterization studies.

justified. To complete the PFMEA, the table includes columns covering information about responsibilities, actions taken and the reassessment (see below).

209

2.5. Computer simulations

210

For the risk evaluation the coating process was modeled with a discrete element method (DEM) simulation and a post-processing algorithm (ray tracing) for the spray simulation. DEM is a simulation technique for granular flows, i.e., it predicts particle movement in particle-processing unit operations. It is based on determining the particle position, the rotational and linear velocities and acceleration at each time step by solving Newton’s equation of motion and a rotational momentum balance, considering all forces acting on each particle, including gravity, Coriolis forces, particle–particle and particle–wall collision, and optional additional interaction forces (e.g., electrostatic and cohesive interactions). Moreover, interactions with a fluid phase can be considered but are irrelevant in many cases (Zhu et al., 2008, 2007). During the last years DEM has been increasingly applied and was further developed (Freireich et al., 2011; Jajcevic et al., 2013; Just et al., 2013b; Sahni and Chaudhuri, 2011; Suzzi et al., 2012, 2010; Toschkoff et al., 2013). One of the great advantages of DEM is the possibility to monitor each tablet in the batch. Yet DEM has limitations: the number of particles that can be simulated within a reasonable time and the complexity of the applied models depend on the available hardware (Kremer and Hancock, 2006). However, the rapid advances in computer hardware development continuously eliminate these limitations. Current efforts aim to develop a DEM software that uses the capabilities of modern graphics processing units to simulate large numbers of particles in the order of 10–100  106 particles (Jajcevic et al., 2013; Radeke et al., 2010). In this work, EDEM (DEM Solutions Ltd.) was used to analyze the tablet coating process. For the treatment of the particle (tablet) collisions, a soft-sphere approach was applied and the contact force calculation was performed using the Hertz-Mindlin model (Cundall and Strack, 1979; Hertz, 1881). To produce reliable results, a DEM simulation needs a reasonable choice of material properties. They were obtained from previous measurements available in the literature (Just et al., 2013b; Ketterhagen et al., 2010). All material properties, simulation parameters and PP characteristics are listed in Tables 3 and 4 .

211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247

Table 3 List of simulation parameter settings. Parameter

Value

Tablet diameter Tablet density Tablet poisson ratio Coefficient of restitution (tablet–tablet and tablet–wall) Friction (tablet–tablet) Friction (tablet–wall) Rolling friction Simulation time step

8.5 mm 1395 kg/m3 0.29 0.78 0.5 0.15 0.01 4  105 s

The coating process simulation was performed for a laboratory pan coater (BFC 5, L.B. Bohle Maschinen + Verfahren GmbH, Ennigerloh, Germany). The geometry of the coating drum was imported into the software based on the data provided by the manufacturer. Round biconvex tablets with a diameter of 8.5 mm diameter were used, whose shape was approximated via the gluedsphere approach (i.e., seven spheres were fixed to each other to approximate the bi-convex shape) (Kodam et al., 2009). For each set of parameters, the simulation was run for 66 s. The first 6 s were allocated to fill the defined amount of tablets into the rotating coating drum and to initialize a stable tablet motion. The remaining 60 s were the processing time. A time span of 60 s is enough to capture the significant properties of the process, and can therefore be used to predict the final outcome of the coating process if no changes are made on the process conditions (Kalbag and Wassgren, 2009). A snapshot of the simulated coating drum after 2 s of coating is shown in Fig. 3. A basic DEM simulation provides the tablets’ trajectories but not the coating mass the tablets receive. To obtain the coating mass per tablet, an additional method that models the spray was included (Toschkoff et al., 2013). The spray simulations were performed via a post-processing algorithm based on ray-tracing. In this method, the DEM simulations were run without including the spray, and the tablets’ position data were written to a file every 20 ms and then read by the spray algorithm (based on ray-tracing), together with the spray parameters. Two spray nozzles were used, represented in the simulation by their elliptical spray zones. The placement of the nozzles relative to the pan coater is shown in Fig. 4. On this basis, the coating mass of all individual tablets was determined.

Table 4 List of process characteristics implemented in the computer simulation model. Process characteristics

Level or range (fixed process settings for computer simulation)

Coating time Process drying air

60 s Inlet and outlet air flow rate. Assumption for computer simulation: spray is not affected by the air flow. Dependent on the coating time. Is determined experimentally for the nozzle type Schlick 930 ABC, and then implemented in the simulation model. Is retrospectively chosen (assumed to be around 6 g/min for each spray nozzle).

Coating level Spray pattern Spray rate

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277

G Model

IJP 14297 1–11 6

E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

Fig. 3. DEM simulation of tablets in the coating drum after 2 s of process time. Note that while here the current coloring of tablets is shown for guidance, the actual coating mass was determined in post-processing as described in the text. 278

2.6. Design of experiment (DoE) for process parameter screening

279

Process parameter screening studies were planned with the DoE software Mode 9.0 (Umetrics, Sweden) to create and analyze the experimental designs for the factors shown in Table 5. An interaction DoE model based on a full factorial design was performed consisting of 108 runs and one replication at the center point. It is assumed that the factors pan speed, fill level, spray angle, distance nozzle to bed, and spray nozzle position in

280 281 282 283 284 285

286

the pan have a critical influence on the coating mass uniformity. Two pan speed levels of 16 rpm and 20 rpm were set. For the simulation it was assumed that the laboratory pan coater is equipped with two spray nozzles. Fill level was varied between 3.5 kg and 4 kg of tablets. The angles varied between 40 , 45 or 50 . For the settings used here, an angle of 45 meant that the nozzles were oriented perpendicular to the tablet bed surface; for the cases 40 and 50 the nozzle is rotated by 5 around the cylinder axis. The distances between the nozzle and the tablet bed

Fig. 4. Front and top view of the pan coater, indicating the variation in the nozzle position, angle and distance spray gun to the tablet bed for an elliptical spray zone. (For interpretation of the references to color in text, the reader is referred to the web version of this article.)

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

287 288 289 290 291 292 293 294

Q4

G Model

IJP 14297 1–11 E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

7

Table 5 Design of experiment – factor definition.

295 296 297 298 299 300 301 302 303 304 305

Factor name

Abbreviation

Unit

Factor type/use

Settings

Pan speed Fill level Spray nozzle angle Distance nozzle to tablet bed Spray nozzle position

Speed Fill Angle Dist Pos

rpm kg

Multilevel/controlled Multilevel/controlled Multilevel/controlled Multilevel/controlled Multilevel/controlled

16; 20 3.5; 4 40; 45; 50 0.075; 0.1; 0.125 0.5; 0; 0.5

m rel.

were 0.075 m, 0.1 m and 0.125 m. Furthermore, the axial position of the nozzle inside the pan coater was changed. Starting with the standard position, both nozzles were moved axially 33 mm (half the width of the spray) either to the front or the back of the coater (Fig. 4). As DoE response the coefficient of variation (CV (%)) is described as the variation of coating material between the individual tablets in a batch (Turton, 2008). This CQA coating mass uniformity is quantified by the coefficient of variation (CV), which is the ratio of the standard deviation to the mean value of the coating mass, calculated over all tablets: coating mass uniformity. NTab X

1 CV ¼ NTab 306



ðci  cÞ

2

i¼1

c

(1)

311

NTab is the number of tablets in the coater, ci is the coating mass of the tablet i and the mean coating mass cof all tablets. The evaluation of the coating mass uniformity is performed based on the uniformity of dosage units requirements established by the European Pharmacopoeia (European Directorate for the Quality of Medicines, 2013).

312

3. Results and discussion

313

Based on the identified PPs summarized in Fig. 2, an initial risk assessment and a quantitative criticality assessment were performed within a PFMEA. This document was developed by a multidisciplinary team consisting of computer simulation experts, process engineers and a risk management expert. The potential criticality of a PP was classified through risk priority evaluation and integrated in the “PP type” column in Table 6. By means of DoE-based PP screening studies via computer simulation the initial risk assessment was verified. Furthermore, the risk factor severity was quantitatively reassessed.

307 308 309 310

314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341

3.1. Initial risk assessment and quantitative criticality assessment of PPs The PPs fill level and pan speed were assessed concerning their potential failure modes in Table 6 within the “tablet mixing” process phase. It was assumed that the fill level variation has a significant impact on the CV. For this, the fill level of the laboratory pan coater was varied from either being low (3.5 kg per batch) or high (4 kg per batch). Thus, the PP fill level was classified with a moderate to high impact on the CV and can be controlled by a scale before filling the batch into the pan. This leads to classification of a severity level 7, failure occurrence level 3 and detectability of failure as 3. With an RPN of 63, which was below the critical level of 125, the fill level was classified as a GPP and was investigated within DoE-based PP screening studies for batch sizes of 3.5 kg and 4 kg to verify the initial assessment of severity and criticality. Potential failure modes for pan speed were set as low pan speed with 16 rpm and high pan speed with 20 rpm. As the PP can be monitored online, a potential cause or mechanism of failure could

be an operator failure. Furthermore, the required current control measures for preventing occurrence of failure are defined within operator training for the standard operating procedure (SOP). In addition to this, supporting sections in the manufacturing record should instruct the operator and prevent failure occurrence as described in Fig. 5. This led to an initial assessment of the pan speed as a GPP with an occurrence level of 3 and a detectability level of 1. Furthermore, the PP was determined to have a moderate to high impact on CV which was represented by a severity level of 7. Within the “coating” process phase, the PPs spray nozzle position, spray nozzle angle and distance of nozzle to tablet bed were assessed concerning their potential failure modes in Table 6. For the spray nozzle position, a change in axial positive or negative direction was defined as a potential failure mode. The potential failure effect would include a spray loss which, furthermore, leads to a coating mass uniformity not fulfilling the product quality specifications. Operator failure was defined as an example for a potential cause or mechanism of failure within the initial risk assessment for all investigated PPs of the coating phase. Required current controls for preventing failure occurrence are defined within supporting sections in the manufacturing record in Table 6. Especially, for spray nozzle position and angle, an operator failure can be prevented by using a position indicator on the nozzle arm to set the spray nozzle angle and position properly. An appropriate control to detect a failure would be an operator check based on the principle of dual control. Thus, a risk factor level of 5 was chosen for severity, occurrence and detectability based on the risk assessment catalog in Table 1. A change of the PP spray nozzle angle was defined as another potential failure mode. The effect of the identified failure mode can either be spray loss or a critical variation of coating mass uniformity. For prevention of an operator failure, a position indicator on the nozzle arm to set the spray nozzle angle and position can be used. But the failure could also be detected by a second person controlling the angle adjustment. Thus, severity of failure was assessed with level 7, occurrence and detectability with level 5. For the coating process, the distance between nozzle and tablet bed includes potential failure modes distance too low or too high. The variation of nozzle distance to tablet bed was assessed to have moderate to high impact on coating uniformity representing a severity level of 5 and 7. Especially, the distance reduction can cause tablet overwetting, and therefore, a visual check was required for parameter control. This case was assessed with a severity level of 7 and with occurrence and detectability level of 5. Recommended actions were defined within the initial risk assessment, which involve the investigation of impacts on CV via DoE-based process parameter screening studies.

342

3.2. DoE-based PP screening studies via computer simulation

390

After processing the experimental design, including 109 simulation runs, the effect plot (Umetrics Mode, Sweden) was used for model interpretation, as shown in Fig. 4. The effect plot allows a review of the model fit and classifies the importance of

391

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

392 393 394

Verification of initial risk assessment by a quantitative reassessment of severity Recommended actions RPN PP Occur Current controls Detect Current type controls to to prevent detect failure failure occurrence

Fill level too low Coating mass (3.5 kg per batch) uniformity out of specification

7

Operator failure

3

SOP training, manufacturing record

3

Weight control with balance

63

GPP

Fill level (kg)

Fill level too high Coating mass (4 kg per batch) uniformity out of specification

7

Operator failure

3

SOP training, manufacturing record

3

Weight control with balance

63

GPP

Tablet mixing

Pan speed (rpm)

Pan speed low (16 rpm)

Coating mass uniformity out of specification

7

Operator failure

5

SOP training, manufacturing record

1

Online pan speed monitoring

35

GPP

Tablet mixing

Pan speed (rpm)

Pan speed high (20 rpm)

Coating quality not achieved, wrong tablet bed flow

7

operator failure

5

SOP training, manufacturing record

1

Online pan speed monitoring

35

GPP

Coating

Spray nozzle position

Coating mass uniformity out of specification

5

Operator failure

5

CPP

Operator failure

5

125

CPP

Coating

Spray nozzle angle

5 Coating mass uniformity out of specification, spray loss Change spray pattern 7 – influence on CV, spray loss

Operator failure

5

175

CPP

Coating

Spray nozzle angle

Change of angle 5

Operator failure

5

175

CPP

Coating

Distance of nozzle to tablet bed

Operator failure

5

5

175

CPP

Coating

Distance of nozzle to tablet bed

Nozzle distance too low (2,5 cm too close to bed surface) Nozzle distance too high (2.5 cm too far from bed surface)

Change spray pattern 7 –influence on coating mass uniformity, spray loss 7 Coating mass uniformity out of specification, overwetting 7 Coating mass uniformity out of specification

Principle of dual control (four-eye principle) Principle of dual control (four-eye principle) Principle of dual control (four-eye principle) Principle of dual control (four-eye principle) Visual check

125

Spray nozzle position

Manufacturing record, position indicator on nozzle arm Manufacturing record, position indicator on nozzle arm Manufacturing record, position indicator on nozzle arm Manufacturing record, position indicator on nozzle arm Manufacturing record

5

Coating

Position change in axial direction + 0.5 of spray zone width Postition change in axial direction 0.5 of spray zone width Change of angle +5

Operator failure

5

5

Visual check

175

CPP

Potential failure Process-/ equipment- mode design parameter

Tablet mixing

Fill level (kg)

Tablet mixing

Potential failure effect

Manufacturing record

5

5

5

Action: DEM-based computer simulation investigate influence of low fill level variation on CV. Action: DEM-based computer simulation investigate influence of high fill level variation on CV. Action: DEM-based computer simulation investigate influence of pan speed variation on CV. Action: DEM-based computer simulation investigate influence of pan speed variation on CV. Action: computer simulation investigate influence of spray nozzle position variation on CV Action: computer simulation investigate influence of spray nozzle position variation on CV Action: computer simulation investigate influence of spray nozzle angle variation on CV Action: computer simulation investigate influence of spray nozzle angle variation on CV Action: computer simulation investigate distance of nozzle to tablet bed variation on CV Action: computer simulation investigate distance of nozzle to tablet bed variation on CV

Actions taken

Sev RPN

Computer simulation based on process parameter screening DoE. Computer simulation based on process parameter screening DoE. Computer simulation based on process parameter screening DoE. Computer simulation based on process parameter screening DoE. Computer simulation based on process parameter screening DoE Computer simulation based on process parameter screening DoE Computer simulation based on process parameter screening DoE Computer simulation based on process parameter screening DoE Computer simulation based on process parameter screening DoE Computer simulation based on process parameter screening DoE

5

45

5

45

5

25

5

25

7

175

7

175

3

75

3

75

7

175

7

175

E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

Sev Potential cause or mechanism of failure

Process phase

G Model

Initial risk assessment

IJP 14297 1–11

8

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

Table 6 Initial risk assessment followed by verification of the risk factor severity and a criticality reevaluation of process parameters: for each process parameter further investigations based on the DEM computer simulations were performed. The risk priority numbers of the initial risk assessment were compared with the values of the reassessment.

395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451

G Model

IJP 14297 1–11 E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

9

Fig. 5. Effect plot – effects of the spray nozzle position (Pos), distance of nozzle to tablet bed (dist), pan speed (speed), the interaction between fill level and spray nozzle position, spray nozzle angle (angle) and fill level (fill) factors on the coefficient of variation (%).

452 461 460 459 458 457 456 455 454 453 462 463 464 465 466 467 468

each factor, including also interactions. Furthermore, each effect was ranked concerning its magnitude (Fig. 4). Here, the factors listed in Table 5 were plotted in descending order, depending on their absolute value of effect on the coefficient of variation [CV %], which was computed as twice the multiple linear regression (MLR) coefficient. The number of simulation runs, the residual standard deviation (RSD) of 1.654% and the goodness of fit represented by a R2 of 0.962 and a Q2 of 0.957 are illustrated in the summary table of

Fig. 4. For each factor, a confidence interval of 95% was displayed with error bars. As can be seen from Fig. 4, the spray nozzle position has the highest effect on the CV during the coating process phase. Then, the distance of nozzle to tablet bed, pan speed, and the interaction of fill level and spray nozzle position were important as well. Fill level and angle has a minor impact on CV. Interactions, except for fill level with position, were not relevant.

Fig. 6. Modde response 4D contour plots for pan speed at 20 rpm: the inner factor axis include spray nozzle position and fill level. The outer factor axis includes spray nozzle angle and distance of nozzle to tablet bed.

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

469 470 471 472 473 474 475 476

G Model

IJP 14297 1–11 10 477 478 479

E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx

3.3. Verification of the initial assessment of risk and process parameter criticality by a quantitative reassessment of severity (PFMEA)

536

The output of the PP screening studies, illustrated in the Mode response 4D contour plot, was used for verification (Fig. 5). Generally, the 4D axes are split in the inner factor axis including spray nozzle position and fill level, and the outer factor axis including spray nozzle angle and distance of nozzle to tablet bed. The fifth factor pan speed was kept constant at 20 rpm and 16 rpm. However, only the results for a pan speed of 20 rpm are presented as the figure for 16 rpm would be very similar. In Fig. 4, the CV variation was coded by contour colors from red to green. Green to yellow levels indicate low CV values, which represent combinations of factors leading to acceptable coating mass uniformity. Orange to red levels indicate high CV values, which represent combinations of factors leading to poor coating mass uniformity. In summary, optimal CV values were obtained for high fill levels, spray angles of 40 or 45 , a distance of nozzle to tablet bed of 0.125 m and the +0.5 setting for spray zone position. While normally lower fill levels lead to lower CV values, computer simulation results of this study demonstrated lower CV values for higher fill levels, which is a characteristic of this specific pan coater geometry. The quantitative evaluation of the investigated effects on the CV enabled a verification of the initial risk assessment. Results are shown in Table 6. For each PP the reassessment of the risk factor was performed separately, which either resulted in a severity upgrade or downgrade. The PP fill level had a low impact on the CV as shown in Fig. 5, especially for the +0.5 position as seen in Fig. 6. The reassessment of the risk factor severity therefore resulted in a downgrade of severity to level 5, and the RPN value of 45 stated fill level as a GPP for this specific case. Similarly, the severity level for pan speed was downgraded from 7 to 5, because pan speed again is a PP which demonstrated a moderately severe impact on the CV values. Especially, a lower pan speed led to lower CV values, while a higher pan speed led to comparatively higher CV values. The reassessment confirmed the initial criticality classification. In addition, the RPN value was reduced from 35 to 25. In contrast, the spray nozzle position demonstrates higher criticality than initially assumed. Therefore, the severity ranking was upgraded from 5 to 7 for a position change in the positive and negative axial direction, which resulted in an RPN of 175. Here, the consequence was an increase of the RPN value from 125 in the initial risk assessment to 175 in the reassessment, which confirmed the criticality of this PP. Within the initial risk assessment in Table 6, the equipment design parameter spray nozzle angle was assumed to be critical, which resulted in a severity ranking of 7 and an RPN of 175. According to the computer simulation results, a comparatively low to moderate impact on CV values was demonstrated within the investigated angle variation range. Therefore, the severity ranking was downgraded from 7 to 3. This resulted in a reduction of the RPN values from 175 to 75 (Table 6). Thus, the criticality reclassification of the spray nozzle angle was downgraded to a GPP. Finally, the simulation results demonstrated a moderate to high impact on the CV value for the PP distance of nozzle to tablet bed. A reduction of the distance resulted in an increase of the CV value in comparison to the standard distance. The reassessment maintained the severity level of 7, reflecting the importance of preventive and control measures (Table 6).

537

4. Summary and conclusions

480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535

538 539

This study demonstrates the usefulness of computer simulation as a quantitative tool for risk ranking to improve objectivity in

quality risk management. QRM clearly benefits from applying computer simulation to gain a higher level of objectivity. Moreover, simulations can guide experimental studies, especially concerning scale-up and control. This work specifically highlights the potential for an active-film coating process. According to the PP screening studies, factors like distance of nozzle to tablet bed and spray nozzle position demonstrated a significant impact on the coefficient of variation. Especially the spray nozzle position had the highest influence on the CV, followed by the distance of the nozzle to the tablet bed, pan speed, and the interaction of fill level and spray nozzle position. According to the results of the PP screening studies, the initial severity assessment was adapted in the PFMEA. Here, especially, the spray nozzle position demonstrated higher impact on the CQA than initially assumed. The severity ranking for the spray nozzle angle and pan speed was downgraded. Furthermore, the criticality classification of PPs verified correspondence with the initial criticality assessment. In this work, the major advantages of computer simulations, such as a rapid variation of the parameters, was demonstrated. This can be seen as a large benefit especially for large series of screening experiments. The generated process understanding can lead to a significant reduction of the required number of empirical experiments. Next steps would involve actual experiments to validate the computer simulation results. Moreover, a design space within the knowledge space could be established based on the generated experimental data. This allows an optimization of the design space development, reducing process development time and costs. Furthermore, the novel approach described in this study clearly demonstrates the potential of simulation within the Quality-byDesign framework and will enable more objectivity within assessment, control, communication and review of quality risks of pharmaceutical manufacturing processes.

540

Acknowledgements

573

This work has been funded by the Austrian COMET Program under the auspices of the Austrian Federal Ministry of Transport, Innovation and Technology (bmvit), the Austrian Federal Ministry of Economy, Family and Youth (bmwfj) and by the State of Styria (Styrian Funding Agency SFG). COMET is managed by the Austrian Research Promotion Agency FFG. The authors thank Otto Scheibelhofer for his helpful comments.

541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572

Q2 574 575 576 577 578 579 580

References

581

Aksu, B., De Beer, T., Folestad, S., Ketolainen, J., Lindén, H., Lopes, J.A., de Matas, M., Oostra, W., Rantanen, J., Weimer, M., 2012. Strategic funding priorities in the pharmaceutical sciences allied to quality by design (QbD) and process analytical technology (PAT). Eur. J. Pharm Sci. 47, 402–405. doi:http://dx.doi.org/10.1016/j. ejps.2012.06.009. Chen, W., Chang, S.-Y., Kiang, S., Marchut, A., Lyngberg, O., Wang, J., Rao, V., Desai, D., Stamato, H., Early, W., 2010. Modeling of pan coating processes: prediction of tablet content uniformity and determination of critical process parameters. J. Pharm Sci. 99, 3213–3225. doi:http://dx.doi.org/10.1002/jps.22044. CMC Biotech Working Group, 2009. A-Mab: a case study in bioprocess development (WWW Document). http://www.casss.org/associations/9165/files/A-Mab_Case_Study_Version_2-1.pdf (accessed 7.15.14). Cole, G., Hogan, J., Aulton, M., 1995. Pharmaceutical Science Series: Pharmaceutical Coating Technology. Taylor & Francis Ltd., London. Cundall, P.A., Strack, O.D.L., 1979. A discrete numerical model for granular assemblies. Geotechnique 29, 47–65. doi:http://dx.doi.org/10.1680/ geot.1979.29.1.47. Dubey, A., Hsia, R., Saranteas, K., Brone, D., Misra, T., Muzzio, F.J., 2011. Effect of speed, loading and spray pattern on coating variability in a pan coater. Chem. Eng. Sci. 66, 5107–5115. doi:http://dx.doi.org/10.1016/j.ces.2011.07.010. European Directorate for the Quality of Medicines, 2013. European Pharmacopoeia 7.8, 7th ed. . European Medicines Agency, 2009. Pharmaceutical Development Q8(R2). European Medicines Agency, 2011. Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities) Q11.

582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

Q3 602 603 604 605

G Model

IJP 14297 1–11 E. Stocker et al. / International Journal of Pharmaceutics xxx (2014) xxx–xxx 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657

Freireich, B., Ketterhagen, W.R., Wassgren, C., 2011. Intra-tablet coating variability for several pharmaceutical tablet shapes. Chem. Eng. Sci. 66, 2535–2544. doi: http://dx.doi.org/10.1016/j.ces.2011.02.052. Hertz, H., 1881. Über die Berührung fester elastischer Körper. J. Reine Angew. Math 92, 156–171. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 1999. ICH Harmonised Tripartite Guideline Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances Q6A. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2004. Organisation of the Common Technical Document for the Registration of Pharmaceuticals for Human Use M4 Step 4 (R3) Version. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005. ICH Harmonised Tripartite Guideline Quality Risk Management Q9. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2008. ICH Harmonised Tripartite Guideline Pharmaceutical Quality System Q10. International Electrotechnical Commission IEC 60812, 2006. Analysis Techniques for System Reliability- Procedures for Failure Mode and Effects Analysis (FMEA). International Organization for Standardization, 2007. ISO 14971:2007 – Application of Risk Management to Medical Devices. International Organization for Standardization, 2009a. ISO/FDIS 31010 2009 Risk Management – Risk Assessment Techniques ISOIEC 31,010. ISO/IEC. International Organization for Standardization, 2009b. ISO Guide 73 Risk Management—Vocabulary ISO Standard. Jajcevic, D., Siegmann, E., Radeke, C.A., Khinast, G.J., 2013. Large-scale CFD-DEM simulations of fluidized granular systems. Chem. Eng. Sci. 98, 298–310. doi: http://dx.doi.org/10.1016/j.ces.2013.05.014. Just, S., Toschkoff, G., Funke, A., Djuric, D., Scharrer, G., Khinast, J., Knop, K., Kleinebudde, P., 2013a. Optimization of the inter-tablet coating uniformity for an active coating process at lab and pilot scale. Int. J. Pharm. 457, 1–8. doi:http:// dx.doi.org/10.1016/j.ijpharm.2013.09.010. Just, S., Toschkoff, G., Funke, A., Djuric, D., Scharrer, G., Khinast, J.G., Knop, K., Kleinebudde, P., 2013b. Experimental analysis of tablet properties for discrete element modeling of an active coating process. Am. Assoc. Pham. Sci. J. Pharm. Sci. Technol. 14, 402–411. doi:http://dx.doi.org/10.1208/s12249-013-9925-5. Kalbag, A., Wassgren, C., 2009. Inter-tablet coating variability: tablet residence time variability. Chem. Eng. Sci. 64, 2705–2717. doi:http://dx.doi.org/10.1016/j. ces.2009.02.037. Kalbag, A., Wassgren, C., Sumana Penumetcha, S., Pérez-Ramos, J.D., 2008. Intertablet coating variability: residence times in a horizontal pan coater. Chem. Eng. Sci. 63, 2881–2894. doi:http://dx.doi.org/10.1016/j.ces.2008.03.009. Kessel, M., 2011. The problems with today's pharmaceutical business-an outsider’s view. Nat. Biotechnol. 29, 27–33. Ketterhagen, W.R., Bharadwaj, R., Hancock, B.C., 2010. The coefficient of rolling resistance (CoRR) of some pharmaceutical tablets. Int. J. Pharm. 392, 107–110. doi:http://dx.doi.org/10.1016/j.ijpharm.2010.03.039. Kodam, M., Bharadwaj, R., Curtis, J., Hancock, B., Wassgren, C., 2009. Force model considerations for glued-sphere discrete element method simulations. Chem. Eng. Sci. 64, 3466–3475. doi:http://dx.doi.org/10.1016/j.ces.2009.04.025. Kremer, D.M., Hancock, B.C., 2006. Process simulation in the pharmaceutical industry: a review of some basic physical models. J. Pharm. Sci. 95, 517–529. doi: http://dx.doi.org/10.1002/jps.20583.

11

Liu, H.-C., Liu, L., Liu, N., 2013. Risk evaluation approaches in failure mode and effects analysis: a literature review. Expert Syst. Appl. 40, 828–838. Milá, L., Valdés, R., Tamayo, A., Padilla, S., Ferro, W., 2012. Application of a risk analysis method to different technologies for producing a monoclonal antibody employed in hepatitis B vaccine manufacturing. Biologicals 40, 118–128. doi: http://dx.doi.org/10.1016/j.biologicals.2011.12.004. Porter, S.C., Verseput, R.P., Cunningham, C.R., 1997. Process optimization using design of experiments. Pharm. Technol. 21, 60–70. Radeke, C.A., Glasser, B.J., Khinast, J.G., 2010. Large-scale powder mixer simulations using massively parallel GPU architectures. Chem. Eng. Sci. 65, 6435–6442. doi: http://dx.doi.org/10.1016/j.ces.2010.09.035. Sahni, E., Chaudhuri, B., 2011. Experiments and numerical modeling to estimate the coating variability in a pan coater. Int. J. Pharm. 418, 286–296. doi:http://dx.doi. org/10.1016/j.ijpharm.2011.05.041. Stamatis, D.H., 2003. In: William, A. (Ed.), Failure Mode Effects Analysis: FMEA from Theory to Execution. 2nd ed. Tony, Milwaukee. Suzzi, D., Radl, S., Khinast, J.G., 2010. Local analysis of the tablet coating process: impact of operation conditions on film quality. Chem. Eng. Sci. 65, 5699–5715. doi:http://dx.doi.org/10.1016/j.ces.2010.07.007. Suzzi, D., Toschkoff, G., Radl, S., Machold, D., Fraser, S.D., Glasser, B.J., Khinast, J.G., 2012. DEM simulation of continuous tablet coating: effects of tablet shape and fill level on inter-tablet coating variability. Chem. Eng. Sci. 69, 107–121. doi: http://dx.doi.org/10.1016/j.ces.2011.10.009. Tobiska, S., Kleinebudde, P., 2003. Coating uniformity and coating efficiency in a Bohle Lab-Coater using oval tablets. Eur. J. Pharm. Biopharm. 56, 3–9. doi:http://dx.doi.org/10.1016/S0939-6411(03)00026-2. Toschkoff, G., Just, S., Funke, A., Djuric, D., Knop, K., Kleinebudde, P., Scharrer, G., Khinast, J.G., 2013. Spray models for discrete element simulations of particle coating processes. Chem. Eng. Sci. 101, 603–614. doi:http://dx.doi.org/10.1016/j. ces.2013.06.051. Toschkoff, G., Khinast, J.G., 2013. Mathematical modeling of the coating process. Int. J. Pharm. 457, 407–422. doi:http://dx.doi.org/10.1016/j.ijpharm.2013.08.022. Toschkoff, G., Suzzi, D., Khinast, J.G., Tritthart, W., Reiter, F., Schlingmann, M., 2012. Detailed analysis of air flow and spray loss in a pharmaceutical coating process. AIChE J. 58, 399–411. doi:http://dx.doi.org/10.1002/aic.12681. Turton, R., 2008. Challenges in the modeling and prediction of coating of pharmaceutical dosage forms. Powder Technol. 181, 186–194. doi:http://dx.doi.org/10.1016/j.powtec.2006.12.006. World Health Organization, 2010. Guideline on quality risk management working document QAS/10.376 (WWW Document). http://www.who.int/medicines/ services/expertcommittees/pharmprep/QualityRiskManagement-QAS10376_18082010.pdf (accessed 7.15.14). Zhu, H.P., Zhou, Z.Y., Yang, R.Y., Yu, A.B., 2008. Discrete particle simulation of particulate systems: a review of major applications and findings. Chem. Eng. Sci. 63, 5728–5770. doi:http://dx.doi.org/10.1016/j.ces.2008.08.006. Zhu, P., Zhou, Z.Y., Yang, R.Y., Yu, A.B., 2007. Discrete particle simulation of particulate systems: theoretical developments. Chem. Eng. Sci. 62, 3378–3396. doi:http://dx.doi.org/10.1016/j.ces.2006.12.089. Zimmermann, H.F., Hentschel, N., 2011. Proposal on how to conduct a biopharmaceutical process failure mode and effect analysis (FMEA) as a risk assessment tool. PDA J. Pharm. Sci. Technol. 65, 506–512. doi:http://dx.doi.org/10.5731/ pdajpst.2011.00784.

Please cite this article in press as: Stocker, E., et al., Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework. Int J Pharmaceut (2014), http://dx.doi.org/10.1016/j.ijpharm.2014.08.055

658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704

Use of mechanistic simulations as a quantitative risk-ranking tool within the quality by design framework.

The purpose of this study is to evaluate the use of computer simulations for generating quantitative knowledge as a basis for risk ranking and mechani...
2MB Sizes 1 Downloads 3 Views