(621c) Data Reconciliation in the Quality-By-Design (QbD) Implementation of Pharmaceutical Continuous Tablet Manufacturing

Authors: 
Su, Q., Purdue University
Bommireddy, Y., Purdue University
Gonzalez, M., Purdue University
Reklaitis, G. V., Purdue University
Nagy, Z. K., Purdue University
Pharmaceutical continuous manufacturing is now entering the arena of big data with more and more critical quality attributes (CQA) and critical process parameters (CPP) readily available in real time using process analytical technology (PAT). Although the management and integration of process operational and product quality data have been receiving extensive interests,1 the transition from data-rich to information and knowledge-rich operation has not received much attention thus far.2 Under the conventional implementation of multivariate statistical process control (MSPC), the nominal variation in the process systems (equipment\operation\material) and sensor measurements are monitored by tracking a small number of principal components and testing them against pre-specified statistical control limits, e.g., Hoteling’s T2 and square prediction error (SPE). However, while simplifying monitoring, this approach does not directly use the valuable information on the true product quality variations, which are more directly-related to in-process and final product quality attributes.3 Furthermore, there are some critical material attribute (CMA) or CQA variables that cannot be measured in line and thus cannot be captured by MSPC methods. Finally, an adequate database of nominal operations for MSPC development is not always available, especially not during the early-stages of product/process development.

Data reconciliation (DR) is an important process systems engineering tool for accommodating process and measurement variations and estimating the true state of the manufacturing system. Based on process mathematical models, DR seeks to minimize the errors between model predicted system states and the state information provided by sensor measurements that inherently have different levels of accuracy and precision. Effectively by reconciling the model predictions of system states with the nominal sensor measurements, the differences in sensor measurement errors and offsets are taken into account. This is in consistent with the US FDA Quality by Design (QbD) guidelines, which highlights the importance of process understanding in process development and product quality control. The role of DR is especially important given the significant differences which normally occur between CQA and CPP variable variations. For example, most CQA variables are now indirectly measured using in situ spectroscopy probes, e.g., active pharmaceutical ingredient (API) composition measured by Near Infrared (NIR) probe. Such measurements are often subject to variations/drifts due to fouling, inherent errors in chemometric model predictions, material property changes (particle size, bulk density, etc.), environmental factor changes (humidity, temperature, etc.), and so on. On the other hand, the CPP variables are commonly and directly measured using reliable mechanical or electrical sensors, e.g., the compression force in a tablet press, which tend to have smaller measurement errors. It is worth mentioning, that the robust design of traditional manufacturing equipment (tablet press, roller compactor, etc.) which provided minimum variation in CPPs, has allowed pharmaceutical manufacturing to continue to operate in batch mode while handling CQA variations via end of line statistical quality control (SQC) methods. The challenge in continuous operation is to effectively integrate the noisier and possibly biased CQA measurements into the process control system so as to effectively supervise the control of CPP’s while minimizing the need for end-on-line SQC. However, this integration does impose additional dynamics on the process, potentially amplifying variations in CPPs and thus in product quality attributes.

In this study, a data reconciliation framework is proposed for implementation in pharmaceutical continuous manufacturing of oral solid dosage. The special features of using the Welsch robust estimator for the errors between model prediction and sensor measurement, 4 and the updating of model parameters to address the model-plant or model-material mismatches are highlighted. The Welsch robust estimator is shown to be capable of rejecting the measurement gross errors without affecting the reconciled model predictions and the robust operation of process control systems. The role of the estimator is especially important when the sensors were undergoing maintenance or replacement due to failures. The real-time model parameter updating is useful when there was a change in raw material properties or process operations. Moreover, the resulting changes in model parameters can be further used to diagnose the system variations in process operation and material properties, as well as to quantify their effects on product quality attributes. The proposed data reconciliation framework is demonstrated on a continuous tablet manufacturing process via direct compaction, which uses a Natoli BLP-16 tablet press. The Kawakita model which captures the relationship between main compression force (CPP) and tablet weight (CQA) is incorporated in the DR framework.5 The Kawakita model parameter are automatically updated to accommodate the changes in material compactibility, e.g., bulk density, particle size, etc. The reconciled main compression force and tablet weight values are used in a hierarchical three-level control design to maintain consistent tablet weight production.6 A tablet weight measurement based on a load cell is used to provide real time but higher variability measurement. The reconciled model predictions of tablet weight are periodically verified and corrected using at-line Sotax Auto Test 4 tablet weight measurements.

The performance of the data reconciliation framework using our developed control system design for continuous tablet manufacturing is quite promising in offering a robust tablet weight control.

References

1.

Markl D, Wahl PR, Menezes JC, et al. Supervisory control system for monitoring a pharmaceutical holt melt extrusion process. AAPS PharmSciTech. 2013;14(3):1034-1044.

2.

Ierapetritou M, Muzzio F, Reklaitis G. Perspectives on the continuous manufacturing of powder-based pharmaceutical processes. AIChE Journal. 2016;62(6):1846-1862.

3.

Moreno M, Liu J, Ganesh S, et al. Steady-state data reconciliation of a direct compression tableting line. Paper presented at: AIChE Annual Meeting, 2017; Minneapolis, US.

4.

Liu J, Su Q, Moreno M, Laird C, Nagy Z, Reklaitis G. Robust state estimation of feeding-blending systems in continuous pharmaceutical manufacturing. Chemical Engineering Research and Design. 2018:Accepted.

5.

Su Q, Bommireddy Y, Gonzalez M, Reklaitis GV, Nagy ZK. Variation and risk analysis in tablet press control for continuous manufacturing of solid dosage via direct compaction. Paper presented at: The 13th International Symposium on Process Systems Engineering PSE 2018, 2018; San Diego, CA.

6.

Su Q, Moreno M, Giridhar A, Reklaitis GV, Nagy ZK. A systematic framework for process control design and risk analysis in continuous pharmaceutical solid-dosage manufacturing. Journal of Pharmaceutical Innovation. 2017;12:327-346.