The emergence of novel pharmaceutical dosage forms and complex technological processes in the beginning of the 21st century has led to rapid development of systems and technological solutions for enhanced control of medicine manufacturing. A system for designing, analysing, and controlling manufacturing through timely measurements (i.e., during processing) of critical material attributes and critical process parameters with the goal of ensuring final product quality is known as process analytical technology or PAT. Different tools, such as statistical and mathematical data management approaches, process analysers, process control tools, and efficient knowledge management tools are used within the PAT framework. The main focus of our doctoral thesis was to evaluate the applicability of different process analysers in the monitoring and control of the pellet film coating process. Firstly, near-infrared (NIR) spectroscopy, which is the most commonly used real-time analyser for monitoring of pellet coating processes, was investigated. In the second part of the thesis, applicability in determination of crucial pellet attributes of another spectroscopic technique, Raman spectroscopy, was evaluated. In addition, two different approaches for measuring particle size parameters during the coating process were examined, i.e. spatial filtering technique (SFT) and image analysis. Multivariate data analysis was used to achieve high usability of the collected process data. The doctoral thesis consists of four articles published in the recognised scientific journals. In the first part of this thesis a review article is presented. The article overviews different options for monitoring of solid dosage forms coating processes in real-time. NIR spectroscopy is the most frequently evaluated approach in the scientific literature, confirming the wide use of this approach for determination of multiple critical attributes during the film coating process of both tablets and pellets. Raman spectroscopy is also often mentioned in the published literature, but unlike NIR spectroscopy, the applicability in monitoring of pellet coating process is less investigated for this technique. Both spectroscopic techniques are considered as ideal PAT analysers, since multiple solid dosage form attributes, such as coating thickness, coating weight, active ingredient content, and drug release rate can be predicted from the spectral data with the use of multivariate models. Theoretically, Raman spectroscopy is less suitable for the measurement of polar molecules than NIR spectroscopy. Therefore, NIR spectroscopy is used for determination of water content during the coating process more commonly. Curing process of the polymer films is poorly investigated in case of both spectroscopic techniques. Second group of process analysers comprises techniques for direct coating thickness determination (terahertz pulse imaging or TPI) or direct particle size measurement (chord length analysers and particle image analysers). Several publications confirm high applicability of these techniques. In addition, all analysers from this group are commercially available and thus commonly used in the manufacturing of solid dosage forms. Besides well-established PAT analysers listed above, some novel approaches have emerged recently, i.e. optical coherence tomography, acoustic emission, microwave resonance, and laser induced breakdown spectroscopy. These approaches have high potential to become valuable PAT tools in the near future. The first research article investigated applicability of the SFT probe in particle size determination and applicability of the NIR spectroscopy in water content and film coating thickness measurement in real-time. The SFT probe proved as an efficient tool for monitoring of the coating process, which was capable of determining the increase in pellet size growth rate after each increase of the coating dispersion spray rate. In addition, measured particle size correlated very well with the reference methods of sieve analysis and static image analysis. The biggest drawback of this measuring approach was its inability to detect agglomeration phenomenon, since the SFT in-line measurements did not correlate with the actual number of agglomerates. The results of the NIR probe evaluation were also interesting. The NIR spectra collected during the pellet coating process correlated very well with the reference methods of loss on drying (LOD) and Karl Fischer titration. Consequently, calibrated multivariate models enabled water content prediction in the pellets with altered qualitative composition, where a different type of the polymethacrylic polymer was used. In addition, the NIR probe proved as a suitable analyser for measuring coating thickness during the coating process. The model, calibrated outside the spectral bands characteristic for water, enabled efficient prediction of the pellet coating thickness when coating of an independent batch was performed. Moreover, the NIR spectral data correlated very well with the reduction of the coating thickness when attrition of the pellets had occurred. The findings listed above illustrate wide applicability of the tested PAT analysers during the coating process. A study on the applicability of the NIR spectroscopy was extended with the second research article, which evaluated the possibility of monitoring the curing process of the prolonged release coating. Unlike the first research article, in which the NIR spectra were recorded in an in-line mode, the spectra of pellets were recorded in an at-line mode in the second article. It was shown that curing of the coated pellets led to decreased peak intensities of the components located below the prolonged release polymer coating. A possible explanation for such results was that increased inter-diffusion of the polymer chains and coalescence of the polymer particles caused reduced intensities of the observed peaks. The NIR spectral data was calibrated with the drug release profiles – measured in three time points – and the calibrated model was later used for the prediction of the active ingredient release rate of the independent sample cluster to evaluate the applicability of the tested approach. The predicted values correlated well with the actual drug release rate with the exception of samples with high water content (higher than 4.5 %). Direct effect of water on spectral data or indirect effect of water on the structure of the polymer film were identified as most probable causes for reduced predictive power of the model in case of the samples with high water content. Moreover, it was confirmed in this study that an at-line NIR spectroscopy is capable to determine the film coating thickness, which additionally confirmed the findings presented in our first research article. The second research article has a high scientific added value, since such findings of the pellet curing process evaluation are presented in no other published work so far. In the last research article the possibility to calibrate the model for in-line monitoring of water content and film coating thickness in real-time by measuring the NIR spectra of laboratory samples off-line was presented. The efficiency of such approach, which can significantly reduce the effort needed to calibrate a model for in-line monitoring, was confirmed by coating thickness and water content prediction of three independent pilot scale batches. In the second part of this article, extensive study on the applicability of Raman spectroscopy in both in-line and at-line mode was presented. The findings of the at-line process analyser evaluation showed that Raman spectroscopy was a very efficient tool for coating thickness determination and, surprisingly, was also capable to assess water content in the tested pellets quite accurately. On the other hand, in-line Raman spectroscopy was not capable to predict water content in the pellets. However, the in-line analyser proved as an efficient tool for accurate coating thickness prediction and amount of sprayed coating dispersion prediction during the pellet coating process. Models calibrated with the off-line Raman spectra were used for the in-line prediction of critical pellet attributes and the results showed that above presented novel approach of calibrating multivariate models could also be applicable in the case of Raman spectroscopy. Finally, a robustness of a novel image analyser was evaluated by monitoring pellet coating process of three pilot scale batches in real-time. Particle size parameters measured in-line were consistent with the course of the process. In addition, measured coating thickness in real-time was compared with the coating thickness measured off-line using dynamic image analysis and no significant difference was observed between both measuring principles. Extensive evaluation of four different process analysers was carried out in the present doctoral thesis. The findings of the thesis are an important extension of the existing knowledge about the PAT tools applicability and they simultaneously present some new aspects that had not been explored previously in the existing scientific publications. The presented knowledge can thus help enhance and more efficiently implement such tools in the development and manufacturing of the medicines. Moreover, it raises some new perspectives, which could be a subject of future studies.
|
