(642c) Two-Dimensional Moisture Content and Size Evolution during Fluid Bed Drying As Part of a Continuous Pharmaceutical Manufacturing Process Using NIR-CI

The pharmaceutical industry is currently making the transition from batch to continuous manufacturing for the latter method’s benefits of e.g. process flexibility, less waste of materials and improved control of the product. For the manufacturing of oral solid dosages, the innovative ConsiGma‑25™ continuous wet granulation line applies twin-screw granulation for production of granules for qualitative tableting. To enable downstream processing, the line uses a semi-continuous fluid bed dryer which reduces the excess liquid in the wet granules coming from the twin-screw wet granulation unit. This process is robust, yet features a multitude of factors that make the dryer system complex. Firstly, the dryer distributes a continuous stream of wet granules over its six drying cells. The cells are sequentially filled for a fixed amount of time, after which the granules are dried and discharged. Due to the gradual filling of each cell, the wet granules start drying at different points in time and the drying regime changes over time. Furthermore, critical process parameters such as the inlet air flow rate, inlet air temperature and cell filling time influence the fluidization and evaporation behavior of the material in the dryer cell. Finally, also the granule size has a major influence on the drying behavior (De Leersnyder et al., 2018).

The study of this complex process, with the aim of Design Space determination in the Quality-by-Design paradigm of the pharmaceutical manufacturing industry (ICH, 2009), is therefore best performed by means of mechanistic modelling. In this way the various interacting phenomena such as fluidization behavior, single particle-level drying and conditions of the drying air can be captured in order to predict the drying behavior of the key product. In models with many interacting factors such as fluid bed drying, sufficiently detailed data is necessary to calibrate and validate the model, to ensure that the final model is the correct representation of the process. As such, these drying models do predict a moisture content distribution in the material (Ghijs et al., 2019; Peglow et al., 2007). These arise from fluidization behavior, the continuous filling of a dryer cell and the heterogeneity in granule size encountered in pharmaceutical granules. Yet in practice, the only available product data is the average moisture content of the granules, meaning that the model cannot compare its output to reality in full extent. This makes that some parts of the model might be insufficiently validated with the reality, and therefore are not able to extrapolate to other process conditions or systems. With two-dimensional moisture content and size distribution data, these models instead would be able to compare their output one-to-one to experimental data, and more components of the model (effect of fluidization, granule size, air conditions) can be validated, which enhances its predictive power.

Therefore, in this study an at-line analytical method was developed to simultaneously determine the moisture content as well as size of individual pharmaceutical granules to construct a 2D distribution. Spatially distributed moisture content measurements of the granules were obtained by dispersing the granules on a conveyor belt (ENP, Hjälteby, Sweden) with a Retsch DR 100 vibrating feeder (Retsch, Haan, Germany), which passed the granules underneath a hyperspectral camera: a near-infrared chemical imaging (NIR-CI) device (VLNIR with OLES56 lens, Specim, Finland). Method validation experiments were performed by measuring multiple samples of granules first with the NIR-CI set-up, immediately followed by a reference LOD instrument (Mettler LP16, Mettler-Toledo, Zaventem, Belgium). By drying wet granules (i) in open air, (ii) in a 50°C oven and (iii) in a fluid bed dryer (Glatt GPCG1), granules with different levels of moisture content were targeted for method validation. All hyperspectral data was then processed in MATLAB (MathWorks®, Natick, Massachusets, USA) using the HYPER-tools package (Mobaraki & Amigo, 2018). The obtained spectra were processed in order to analyze the granules captured on the image. To relate the granules’ spectra to the actual moisture content, the average spectrum of the granules on each image was obtained and compared to the LOD measurement of the sample. A partial least squares (PLS) model was used to derive the moisture content from the spectra. Good model predictions were obtained ), and thus the spectra could accurately indicate the actual moisture content, and this even for granules larger than 3 mm.

In the same setup, the size of each granule on the image is determined, and by means of the PLS-model the moisture content of each granule is found, resulting in a two-dimensional moisture content and size distribution. These distributions capture the spatial distribution of the moisture in the key product. By analyzing the distribution at different drying times and process conditions, highly accurate information is obtained on the evolution of moisture content of the material. In conjunction with a 2D population balance model, this will allow to gain more in-depth knowledge of the drying process. The latter is planned as next step and is out of scope of this abstract which focuses on the data collection.

References

De Leersnyder, F., Vanhoorne, V., Bekaert, H., Vercruysse, J., Ghijs, M., Bostijn, N., ... De Beer, T. (2018). Breakage and drying behaviour of granules in a continuous fluid bed dryer: Influence of process parameters and wet granule transfer. European Journal of Pharmaceutical Sciences, 115, 223–232. http://doi.org/10.1016/j.ejps.2018.01.037

Ghijs, M., Schäfer, E., Kumar, A., Cappuyns, P., Van Assche, I., De Leersnyder, F., ... Nopens, I. (2019). Modeling of Semicontinuous Fluid Bed Drying of Pharmaceutical Granules With Respect to Granule Size. Journal of Pharmaceutical Sciences, 1–8. http://doi.org/10.1016/j.xphs.2019.01.013

ICH. (2009). International Conference on Harmonisation (ICH) of Technical Requirement for Registration of Pharmaceuticals for Human Use, Pharmaceutical Development, Q8 (R2).

Mobaraki, N., & Amigo, J. M. (2018). HYPER-Tools. A graphical user-friendly interface for hyperspectral image analysis. Chemometrics and Intelligent Laboratory Systems, 172(October 2017), 174–187. http://doi.org/10.1016/j.chemolab.2017.11.003

Peglow, M., Kumar, J., Heinrich, S., Warnecke, G., Tsotsas, E., Mörl, L., & Wolf, B. (2007). A generic population balance model for simultaneous agglomeration and drying in fluidized beds. Chemical Engineering Science, 62(1–2), 513–532. http://doi.org/10.1016/j.ces.2006.09.042