(641b) In-Depth Analysis of the Impacts of Material Properties on Size Distributions in Continuous Twin-Screw Wet Granulation to Construct a Generic 1D Population Balance Model

Continuous twin-screw wet granulation (TSWG) is a key unit operation for the continuous manufacturing of several solid oral dosages. To understand the mechanism of TSWG deeply, numerous efforts have been made based on experimental approaches. For example, the liquid-to-solid (L/S) ratio has been identified as a critical process parameter for granule quality attributes¹. The knowledge obtained by experimental studies can help the process development for new formulations qualitatively.

In addition to experimental approaches, mechanistic modeling has been widely performed to predict granule quality attributes as well as to explain the phenomena mathematically. Population balance models (PBMs) are popular methods to compute granule particle size distributions (PSDs) in TSWG. So far, multiple research groups have proposed PBMs with different dimensions and kernels^2–6. One of the biggest challenges of PBMs in TSWG is the applicability for different formulations containing new active pharmaceutical ingredients (APIs). While material properties of APIs are variable, e.g., from hydrophilic to hydrophobic materials, most of the PBM studies have focused on specific materials and are hence not generic.

This study presents an in-depth analysis of the impact of material properties in TSWG to construct a generic 1D PBM. A generic PBM can predict granule PSDs of different formulations and process conditions only by the information of material properties as well as process conditions. As a basis of it, a hybrid model of PBM and partial least squares (PLS) was developed to link material properties with granule PSDs. A compartmental one-dimensional PBM developed by the authors' research groups^{7, 8} was calibrated for ten formulations, including seven different APIs with the same fillers and binders. Afterward, the impact of 34 material properties of the seven different APIs as well as L/S ratio on PBM parameters were trained by developing PLS models for each compartment. The hybrid model, linking the PBM and PLS models, was validated by applying it for four new formulations. A sensitivity analysis was performed to determine critical material properties as well as PBM parameters.

The validation results show that the proposed generic model could predict granule PSDs for new formulations in terms of the maximum mean discrepancies, which represents a statistical distance between two distributions. The maximum mean discrepancies between the simulated and the measured results for 18 experiments were between 0.031 and 0.143. General trends of the impact of material properties on PBM parameters were obtained through the variable importance for the projection of the PLS models. Size and moisture-related material properties affected the first wetting zone, whereas density-related material properties had a high impact on the subsequent kneading zones. From the results of sensitivity analysis, the aggregation efficiency at the wetting zone was found to be the most critical PBM parameter. In addition, analytical results showed that moisture-related properties, e.g., solubility, were key material properties to predict granule PSDs.

In conclusion, the developed model has a potential to reduce experimental work in the process development of new oral solid dosages. The hybrid generic model could predict granule PSDs for new APIs when applying different L/S ratios by characterizing material properties, without any additional experiments. Furthermore, the importance of material properties was prioritized by sensitivity analysis, which decreased the number of the necessary material characterization methods for new APIs.

Acknowledgment: The authors would like to acknowledge, in no particular order, Janssen Pharmaceutica, Pfizer Inc., and UCB for their financial support and fruitful collaboration in this project.

References:

1. C. Portier, C. Vervaet, V. Vanhoorne, Continuous twin screw granulation: A review of recent progress and opportunities in formulation and equipment design, Pharmaceutics 13 (2021) 668.

2. K. F. Lee, S. Mosbach, M. Kraft, W. Wagner, A multi-compartment population balance model for high shear granulation, Computers \& Chemical Engineering 75 (2015) 1–13.

3. H. Y. Ismail, M. Singh, A. B. Albadarin, G. M. Walker, Complete two dimensional population balance modelling of wet granulation in twin screw, International Journal of Pharmaceutics 591 (2020) 120018.

4. C. Sampat, Y. Baranwal, R. Ramachandran, Accelerating multidimensional population balance model simulations via a highly scalable framework using GPUs., Computers \& Chemical Engineering (2020) 106935.

5. L. G. Wang, S. U. Pradhan, C. Wassgren, D. Barrasso, D. Slade, J. D. Litster, A breakage kernel for use in population balance modelling of twin screw granulation, Powder Technology 363 (2020) 525–540.

6. L. G. Wang, J. P. Morrissey, D. Barrasso, D. Slade, S. Clifford, G. Reynolds, J. Y. Ooi, J. D. Litster, Model driven design for twin screw granulation using mechanistic-based population balance model, International Journal of Pharmaceutics 607 (2021) 120939

7. D. Van Hauwermeiren, M. Verstraeten, P. Doshi, M. T. am Ende, N. Turnbull, K. Lee, T. De Beer, I. Nopens, On the modelling of granule size distributions in twin-screw wet granulation: Calibration of a novel compartmental population balance model, Powder Technology 341 (2019) 116–125.

8. A. A. Barrera Jiménez, D. Van Hauwermeiren, M. Peeters, T. De Beer, I. Nopens, Improvement of a 1d population balance model for twin-screw wet granulation by using identifiability analysis, Pharmaceutics 13 (2021) 692.