G Model

IJP 14734 1–8 International Journal of Pharmaceutics xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

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

1 2

3 Q1 4 5 6 7 8

Optimization of a pharmaceutical tablet formulation based on a design space approach and using vibrational spectroscopy as PAT tool Pierre-François Chavez a, * , Pierre Lebrun a , Pierre-Yves Sacré a , Charlotte De Bleye a , Lauranne Netchacovitch a , Serge Cuypers b , Jérôme Mantanus b , Henri Motte b , Martin Schubert b , Brigitte Evrard c , Philippe Hubert a , Eric Ziemons a a

University of Liege (ULg), Department of Pharmacy, CIRM, Laboratory of Analytical Chemistry, CHU, B36, 4000 Liege, Belgium UCB Pharma S.A., Avenue de l'Industrie, 1420 Braine-l'Alleud, Belgium c University of Liege (ULg), Department of Pharmacy, CIRM, Laboratory of Pharmaceutical Technology, CHU, B36, 4000 Liege, Belgium b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 22 January 2015 Received in revised form 10 March 2015 Accepted 13 March 2015 Available online xxx

The aim of the present study was to optimize a tablet formulation using a quality by design approach. The selected methodology was based on the variation of the filler grade, taking into account the particle size distribution (PSD) of active pharmaceutical ingredient (API) in order to improve five critical quality attributes (CQAs). Thus, a mixture design of experiments (DoE) was performed at pilot scale. The blending step was monitored using near infrared (NIR) spectroscopy as process analytical technology tool enabling real-time qualitative process monitoring. Furthermore, some tablets were analyzed by Raman imaging to evaluate the API distribution within the samples. Based on the DoE results, design spaces were computed using a risk-based Bayesian predictive approach to provide for each point of the experimental domain the expected probability to get the five CQAs jointly within the specifications in the future. Finally, the optimal conditions of the identified design space were successfully validated. In conclusion, a design space approach supported by NIR and Raman spectroscopy was able to define a blend that complies with the target product profile with a quantified guarantee or risk. ã 2015 Published by Elsevier B.V.

Keywords: Design space Quality by design Process analytical technology Vibrational spectroscopy Optimization Pharmaceutical tablet formulation

9 10 11 12 13 14 15 16 17

1. Introduction In the last few years, the pharmaceutical industries were encouraged by authorities to enhance knowledge and understanding of their products and manufacturing processes. In order to help industries in this task, the International Conference on Harmonization (ICH) has published several guidelines, as the ICH Q8 guideline on pharmaceutical development, which insists on the concept of Quality by design (QbD). In this guideline, QbD is described as a science- and risk-based approach for which the

Abbreviations: AA, average assay; API, active pharmaceutical ingredient; AV, acceptance value; CI, Carr index; CMA, critical material attribute; CQA, critical quality attribute; CU, content uniformity; DHI, distributional homogeneity index; DoE, design of experiments; HR, Hausner ratio; ICH, International Conference on Harmonization; NIR, near infrared; PAT, process analytical technology; PC, principal component; PCA, principal component analysis; PSD, particle size distribution; QbD, quality by design; RSD, relative standard deviation; TPP, target product profile. * Corresponding author at: University of Liege, Department of Pharmacy, Laboratory of Analytical Chemistry, Avenue de l'Hopital 1, CHU, TOUR 4, BAT. B36, +2, 4000 Liège, Belgium. Tel.: +32 4366 4324; fax: number: +32 4366 4317. E-mail address: [email protected] (P.-F. Chavez).

quality should not be tested into products but should be built in by design. Moreover, this QbD approach contributes to a continuous improvement of the product quality by a systematic assessment, understanding and refining of the formulation and processes throughout the product lifecycle (ICH, 2008, 2009). ICH Q8 also defines the design space as “the multidimensional combination and interaction of input variables (e.g., material attributes) and process parameters that have been demonstrated to provide assurance of quality.” (ICH, 2009). It is also mentioned that modifying the operating conditions while staying within the limits of the design space is not considered as a change and does not Q2 require a regulatory post approval change. Process analytical technology (PAT) is described by the Food and Drug Administration (FDA) as “a system for designing, analyzing, and controlling manufacturing through timely measurements (i.e., during processing) of critical quality and performance attributes of raw and in-process materials and processes, with the goal of ensuring final product quality.” (FDA, 2004). Consequently, PAT fits thoroughly with the QbD concept. Near infrared (NIR) spectroscopy is an analytical method that allows acquiring real-time data from manufacturing process without sample preparation or alteration using probes enabling

http://dx.doi.org/10.1016/j.ijpharm.2015.03.025 0378-5173/ ã 2015 Published by Elsevier B.V.

Please cite this article in press as: Chavez, P.-F., et al., Optimization of a pharmaceutical tablet formulation based on a design space approach and using vibrational spectroscopy as PAT tool. Int J Pharmaceut (2015), http://dx.doi.org/10.1016/j.ijpharm.2015.03.025

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

G Model

IJP 14734 1–8 2 40

P.-F. Chavez et al. / International Journal of Pharmaceutics xxx (2015) xxx–xxx

2.3. Tablets manufacturing

101

Tablets were manufactured at pilot scale with a batch size of 7 kg corresponding to 1/10 of the production scale and following the same current instructions of fabrication. The blending process was carried out using a planetary mixer Collette MP20 (GEA Pharma Systems-Collette, Wommelgem, Belgium) and was split into three successive steps. First, the pre-blend was mixed in a 6 L bowl during 15 min afterwards it was mixed with the rest of excipients in a 20 L bowl during 20 min. Finally, the lubricant was added and mixed during 4 min. The mixing speed was set to 69 rpm for all the blending steps. Next, the powder blends were directly compressed using a rotary tablet press R090-F (GEA Pharma Systems-Courtoy, Halle, Belgium) equipped with 18 round punches of 6 mm of diameter and performing a pre-compression before the final compression. The target weight and hardness of tablets were 70 mg and 40 N, respectively. The production output was set to 650 tablets per minute and about 84,000 tablets were produced per batch.

102

2.4. Flowability assessment

119

The flowability assessment of powder was carried out using the tapped density test of the European Pharmacopoeia 2.9.34 to obtain the bulk and the tapped densities (Europe, 2012a). This test was carried out with 100 g samples of each final powder blend introduced in a 250 mL graduated cylinder placed in a tapped density device Stampfvolumeter (JEL, Ludwigshafen, Germany).

120

2.5. PAT tools

126

2.5.1. FT-NIR equipment The blending processes were monitored with a Multi Purpose Analyzer Fourier Transform near infrared spectrometer (Bruker Optics, Ettlingen, Germany) equipped with a thermoelectrically cooled semiconductor Indium Gallium Arsenide (TE-InGaAs) detector. The spectra were collected with a NIR reflectance probe for solids (Bruker Optics) non-invasively interfaced with the blending bowl. Each spectrum was the average of 4 scans and the resolution was set at 16 cm 1 over the range from 12,500 to 4000 cm 1. The time required for a NIR measurement was 1.7 s and time interval between two measures was 1 s. The spectra were collected with the Opus 6.5 software (Bruker Optics).

127

83

in-line measurements (Luypaert et al., 2007). Taking into account its advantages, NIR spectroscopy is a significant tool for the implementation of PAT and has already been described for various applications such as blend homogeneity and coating monitoring or moisture and active content determination (Blanco et al., 2008; Bodson et al., 2007; Grohganz et al., 2010; Krier et al., 2013; Mantanus et al., 2011, 2010; Moes et al., 2008). When considering a tablet manufacturing process, one of the more critical steps is the mixing of the raw materials that should lead to a homogeneous powder blend. Thus, the in-line monitoring of the blending kinetic, using NIR as suitable tool, allows ensuring the quality of the final powder blend. Based on Raman spectroscopy, Raman imaging is an hyperspectral technique providing spectral and spatial information, simultaneously (Gendrin et al., 2008; Sacré et al., 2014b). Therefore, it can be used to obtain distribution maps of compounds of interest (Krier et al., 2013; Mantanus et al., 2011). In this study, the distributional homogeneity index (DHI) developed by Sacré et al. (2014a) has been applied to assess the distributional homogeneity of the active pharmaceutical ingredient (API) in tablets without calibration model. The aim of the present study was to optimize the formulation of a pharmaceutical tablet form which was already manufactured at production scale. Indeed, this tablet has exhibited a high content variation (measured by API content uniformity) with a mean acceptance value (AV) above 10 and an average API assay of 97.8%, biased below the target value of 100%. Thus, it was essential to improve average assay, individual batch Relative standard deviation (RSD) for content uniformity (CU) and AV. In order to stay as close as possible to the current approved registration file, the proposed methodology was based upon the variation of the filler grade, taking into account API particle size distribution (PSD), using a mixture design of experiments (DoE) and a design space approach. Based on the DoE results, design spaces were computed using a risk-based Bayesian predictive approach to provide for each point of the experimental domain the expected probability to get the responses jointly within the specifications in the future runs. This Bayesian predictive approach enables taking into account uncertainties and interactions of the model and is therefore an efficient tool for ensuring the quality as required by ICH Q8 (Castagnoli et al., 2010; Lebrun et al., 2012; Peterson, 2008). As a result, within the framework of a formulation optimization, determining the type and quantity of excipients using a design space approach is fully compliant with the QbD concept.

84

2. Material and methods

140

85

2.1. Tablet formulation

2.5.2. Raman equipment Raman maps were collected with a dispersive Raman spectrometer RamanStation 400 F (PerkinElmer, Waltham, MA, USA) equipped with a two-dimensional CCD detector (1024  256 pixel sensor). The laser excitation wavelength used was 785 nm with a power of 100 mW. An area of 16 mm2 corresponding to the largest square possible was analyzed per tablet with a step size of 100 mm and a map size of 40  40 pixels. Each measure consisted of 1 scan with an exposure time of 1 s and a resolution of 2 cm 1 over the spectral range from 90 to 1622 cm 1. Background measure was repeated each 20 min during mapping. Spectra were collected with Spectrum 6.3.2.0151 software (PerkinElmer). Before analyzing, tablets were prepared to obtain a flat surface with a Leica EM Rapid milling system equipped with a tungsten carbide miller (Leica Microsystems GmbH, Wetzlar, Germany). 2.5.3. Multivariate data treatment NIR spectra were preprocessed using a standard normal variate and a mean centering over the spectral range from 9056 to 4312 cm 1 before being analyzed by principal component analysis (PCA).

156

41 42 43 44 45 46 47 48 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

86 87 88 89 90 91 92 93

The studied formulation was a powder blend of API and excipients for direct compression with a final API concentration of 7.14% (w/w). The main excipient was a filler that was present in the blend at a final concentration of 61.8% (w/w). The remaining Q3 excipients had various functions as binding agent to improve the compatibility of the blend, disintegrant, lubricant and glidant. For confidentiality reasons, no more qualitative or quantitative information about the formulation can be disclosed.

94

2.2. Particle size distribution measurements

95

PSD measurements of API batches were achieved by laser diffraction using a Mastersizer 2000 particle size analyzer (Malvern Instruments, Malvern, UK) equipped with a Scirocco 2000 dry powder dispersion unit (Malvern Instruments). Measurements were realized in triplicates with a dispersion air pressure of 2 bar.

96 97 98 99 100

Please cite this article in press as: Chavez, P.-F., et al., Optimization of a pharmaceutical tablet formulation based on a design space approach and using vibrational spectroscopy as PAT tool. Int J Pharmaceut (2015), http://dx.doi.org/10.1016/j.ijpharm.2015.03.025

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118

121 122 123 124 125

128 129 130 131 132 133 134 135 136 137 138 139

141 142 143 144 145 146 147 148 149 150 151 152 153 154 155

157 158 159

G Model

IJP 14734 1–8 P.-F. Chavez et al. / International Journal of Pharmaceutics xxx (2015) xxx–xxx 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220

PCA is a statistical tool used to compress, extract and ease classification of the information of various data by reducing the number of variables. Thus, the goal of PCA is to represent the data with new variables while keeping the most of information (Burns and Ciurczak, 2008). These new variables, called principal components (PCs), are linear combinations of the original variables and are independent (orthogonal). They explain the largest variability within the data with the first PC summering the most information. The projections of the original data on the PCs are called the scores (Massart et al., 1997). The baseline of Raman spectra was corrected using Asymmetric Least Squares (AsLS) algorithm (Eilers, 2003) with a l value of 105 and a p value of 0.001. Cosmic rays were removed using the algorithm developed by Sabin et al. (2012) with a parameter k of 15 and finally, a mean centering was performed over the whole spectrum. Afterwards, preprocessed data were analyzed by PCA with first PC corresponding to the API spectrum and then, the distribution maps of the API were built by refolding the scores of the first PC. API distribution of Raman maps was investigated using the distributional homogeneity index (DHI) approach with 100 simulations developed by Sacré et al. (2014a). PCA were computed with the PLS Toolbox 7.0.3 (Wenatchee, WA, USA) for Matlab R2013a (The Mathworks, Natick, MA, USA) and DHI were carried out using routines written in Matlab. Experimental data were also treated using Microsoft Excel 2010 (Microsoft, Albuquerque, NM, USA). 2.6. Content uniformity determination CU of ten tablets was determined by HPLC. For confidentiality reasons, this analytical method cannot be disclosed. 2.7. Mixture design of experiments 2.7.1. Critical material attributes Two critical material attributes (CMAs) have been highlighted as having a dramatic effect on the product quality by previous experiments among various parameters that could be modified. These two CMAs are the filler grade and the API PSD. Several filler grades have been tested and three of them were selected. This selection was based on their specific positive influence on the product quality. They have the same chemical composition but different physical properties (PSD, surface area, flowability, etc.). Furthermore, it has been noted that the natural variability of API PSD from batch to batch had an effect on the target product profile. A mixture DoE was developed including the three selected filler grades in ratio from 0.1 to 1 with regard to Filler 1 and from 0 to 0.9 with regard to Fillers 2 and 3 (see Fig. 1). These constrains have been set because Filler 1 is the filler grade currently used in production, and thus its presence in the blend was intended for all experiments. Accordingly, Fillers 2 and 3 can be absent but cannot be the only ones in blend. This DoE also included three API batches with different target PSD. The first API batch was native while the second and the third were modified to obtain larger and smaller particles than native API, respectively (see Section 2.8). Central and extremity points of the DoE were repeated with the three API batches whereas the other points were realized with one of the three API batches randomly selected in a balanced way. The mixture DoE was developed with JMP 10.0.0 software (SAS Institute Inc., Cary, NC, USA) and consisted in 21 experiments. 2.7.2. Critical quality attributes Five critical quality attributes (CQAs) have been analyzed throughout the DoE trials in order to assess the quality of the product. This five CQAs are acceptance value (AV), individual batch

3

Fig. 1. Graph representation of mixture DoE. Central and extremity points are the repeated experiments with the three API batches while other points are experiments performed with API 1 (red color) or API 2 (blue color) or API 3 (green color). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Relative standard deviation (RSD) for content uniformity (CU), average assay (AA), Hausner ratio (HR) and Carr index (CI). Three of them, AV, RSD and AA concern the tablets API content while HR and CI allow the characterization of the flowability of the powder blends. AV is calculated from AA and standard deviation of API content of ten tablets (Europe, 2012c). This value should be lower than 10. The RSD represents the variability of the API contents of the ten tablets and its value should be lower than 3%. AA corresponds to the average of the API content of the ten samples and should be as close as possible to 100%. In the scope of this study, its value should be between 98.5 and 101.5%. HR and CI are obtained based on the bulk and the tapped density of powder. Low HR and CI indicate good flowability properties. Their specifications were defined in order to obtain at least a passable state of flowability (Europe, 2012b). All the specifications have been chosen in a goal of optimization and to meet a minimum satisfactory level of quality. As one can see in Table 1, these defined specifications are more severe than those required by authorities or currently used.

221

2.8. Modification of API particle size distribution

241

As described in Section 2.7.1, the selected approach was to use one API batch with small particles, one with large particles and another with current PSD. To obtain large particles, native API was compacted using a Gerteis Mini-Pactor roller compactor (Gerteis Maschinen + Processengineering AG, Jona, Switzerland). A press force of 10 kN/cm was applied by smooth rollers with a gap of 2 mm and a roller speed of 3 rpm. The compacted API was sieved on a 0.6 mm screen using an oscillating sieving mill Frewitt GLA-ORV-0215 (Frewitt, Granges-Paccot, Switzerland) subsequently. To achieve a small PSD, native API was delumped on a 6H KekGardner Fine Grinding Universal Mill (Kek-Gardner, Macclesfield, UK) equipped with a centrifugal screen of 1 mm with a rotor speed set at 13,000 rpm. The process parameters have been tested upfront to identify the optimal process conditions. Native API was used as current API PSD batch. In order to simplify, API batches have been renamed “API 1” (native), “API 2” (large) and “API 3” (small) throughout this paper.

242

Please cite this article in press as: Chavez, P.-F., et al., Optimization of a pharmaceutical tablet formulation based on a design space approach and using vibrational spectroscopy as PAT tool. Int J Pharmaceut (2015), http://dx.doi.org/10.1016/j.ijpharm.2015.03.025

222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240

243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259

G Model

IJP 14734 1–8 4

P.-F. Chavez et al. / International Journal of Pharmaceutics xxx (2015) xxx–xxx

Table 1 The five CQAs with their defined specifications for optimization and the corresponding specifications required by authorities or currently used (for 10 tablets). CQA

Defined specification for optimization

Specification required by authoritiesa or currently usedb

Acceptance value Relative standard deviation (%) Average assay (%) Hausner ratio Carr index (%)

Optimization of a pharmaceutical tablet formulation based on a design space approach and using vibrational spectroscopy as PAT tool.

The aim of the present study was to optimize a tablet formulation using a quality by design approach. The selected methodology was based on the variat...
1MB Sizes 6 Downloads 12 Views