Combined QbD and Computer Simulation Approaches
[News vom: 28.09.2010]
Quality by Design (QbD) is a systematic science- and risk-based development approach and life cycle management tool that helps to make the right investments in production, research and development in order to characterize increasingly complex processes and to ensure their quality.
The basic concepts of QbD are meanwhile well understood by the scientific community. The fundamental assumption underlying QbD is that if critical sources of variability can be understood, then product performance can be controlled by using the manufacturing process to mitigate variability in the material properties.
Despite the initial hesitation of pharmaceutical industry to adopt QbD methodology, the application of QbD principles has recently received a lot of interest among pharmaceutical companies. As the pharmaceutical industry faces new challenges associated with increased market globalization, higher customer expectations, the push for increased profitability, tighter regulations, and the ever-increasing demand to reduce time to market, the QbD-adaption process will even speed up in near future and will benefit business-, efficacy- and quality goals. Hence, the traditional approach from taking a product from laboratory scale to pilot plant and then to production is no longer attractive. Process and product development are often initiated simultaneously, and as a result, rapid prototyping, e.g., development of formulations, unit operations and associated manufacturing equipment, is required.
Here, computer simulation tools will play a paramount role. For example, process characterization studies for evaluating the impact of potentially critical material attributes and process parameters on critical quality (CQA) and performance attributes could be performed in unprecedented detail. Validated models may directly produce response surfaces and map out the design spaces, where each run of a model may be considered an in silico experiment and surface response generation would be analogous to a DoE without the burden of “physical” experimentation, thus meeting development timelines in a more cost-efficient way. Moreover, sound science data and fundamental process understanding can be generated more efficiently, as not only point data are available but full-field data or data at multiple locations, that are often required to fully characterize a system.
In the pharmaceutical industry, simulation methods used for a detailed process analysis are either computational fluid dynamics (CFD) simulations, or particle-based simulation approaches to study particle flow. CFD simulations are used within the field of bioprocesses, or in spray drying, coating or fluidized bed applications. Particle-based simulations are popular within the powder compaction area, inhalation devices as well as powder blending processes. For example, the discrete element method (DEM) is nowadays widely used within the pharmaceutical industry.
An indispensable prerequisite for the use of simulation tools in a pharmaceutical environment governed by Good Manufacturing Practice (GMP) guidelines is the availability of an efficient validation strategy to prove the predictive capability and robustness of a simulation within a sufficient small confidence interval. Here, the usability of computer simulation based data generation may be twofold. On the one hand, simulations may be used in combination with risk assessment tools at an early stage of process characterization as screening application to map out potentially critical input factors (e.g., raw material characteristics, process parameters, equipment design) and to prioritize them for further “real-life” process characterization. On the other hand, a simulation approach may be used directly in the main phase of process characterization to generate GMP-relevant data that would further be included in the pharmaceutical dossier. In pharmaceutical process development both applications have great potential in streamlining development and manufacturing resources and costs.
Therefore, the RCPE has dedicated noteworthy research resources to develop and validate computer simulation approaches and to challenge their use within the pharmaceutical development community. Hence, combined QbD and computer simulation approaches are performed to characterize key unit operations of solid dosage form manufacturing, namely powder blending and tablet coating. The aim is to evaluate the impact of formulation parameters and process variables on process and product quality. Understanding the variability of both material attributes and process parameters, as well as their overall impact on the unit operations are critical elements for QbD. In a first step, the QbD-methodology is systematically used to
For a blending process the concentration of the active pharmaceutical ingredient, as well as the number of revolutions are related to the blending uniformity. Within the blending simulation approach, several critical points of interest can be effectively addressed by the simulation, e.g. sample locations representing potential areas of poor blending and avoidance of sampling errors. A relationship between low particle number simulated systems and regulatory requirements has been established. By using robust validation strategies, e.g. by the application of PAT, the simulation results can be well correlated to real-life data. A virtual design space is set up, identifying the optimal factor combinations to obtain the best blending output within the given system.
A CFD-simulation approach is used to characterize a spray coating unit operation. Here, the aim is to provide science-based and quantitative understanding of how physicochemical parameters affect the uniformity of the coating layer on a single tablet, passing the spray zone one time at a defined angle. Data then can be used to characterize more complex systems, e.g. tablets passing the spray zone several times at different angles, and knowledge can be leveraged to whole tablet beds. Two potentially critical coating process parameters, i.e. atomizing pressure and inlet temperature, are evaluated and their effect on coating homogeneity is analyzed. By using data provided by the simulation approach, a coating process virtual design space is established. Also here, the optimal combination of input parameters can be identified.
It can be seen that these combined QbD-simulation approaches can well contribute towards an overall pharmaceutical process characterization and highlight several points to consider (e.g. further development of robust validation procedures for models; more efficient simulations that will take into account a higher number of possible input factors) that will be addressed in our future studies.
Here, computer simulation tools will play a paramount role. For example, process characterization studies for evaluating the impact of potentially critical material attributes and process parameters on critical quality (CQA) and performance attributes could be performed in unprecedented detail. Validated models may directly produce response surfaces and map out the design spaces, where each run of a model may be considered an in silico experiment and surface response generation would be analogous to a DoE without the burden of “physical” experimentation, thus meeting development timelines in a more cost-efficient way. Moreover, sound science data and fundamental process understanding can be generated more efficiently, as not only point data are available but full-field data or data at multiple locations, that are often required to fully characterize a system.
In the pharmaceutical industry, simulation methods used for a detailed process analysis are either computational fluid dynamics (CFD) simulations, or particle-based simulation approaches to study particle flow. CFD simulations are used within the field of bioprocesses, or in spray drying, coating or fluidized bed applications. Particle-based simulations are popular within the powder compaction area, inhalation devices as well as powder blending processes. For example, the discrete element method (DEM) is nowadays widely used within the pharmaceutical industry.
An indispensable prerequisite for the use of simulation tools in a pharmaceutical environment governed by Good Manufacturing Practice (GMP) guidelines is the availability of an efficient validation strategy to prove the predictive capability and robustness of a simulation within a sufficient small confidence interval. Here, the usability of computer simulation based data generation may be twofold. On the one hand, simulations may be used in combination with risk assessment tools at an early stage of process characterization as screening application to map out potentially critical input factors (e.g., raw material characteristics, process parameters, equipment design) and to prioritize them for further “real-life” process characterization. On the other hand, a simulation approach may be used directly in the main phase of process characterization to generate GMP-relevant data that would further be included in the pharmaceutical dossier. In pharmaceutical process development both applications have great potential in streamlining development and manufacturing resources and costs.
Therefore, the RCPE has dedicated noteworthy research resources to develop and validate computer simulation approaches and to challenge their use within the pharmaceutical development community. Hence, combined QbD and computer simulation approaches are performed to characterize key unit operations of solid dosage form manufacturing, namely powder blending and tablet coating. The aim is to evaluate the impact of formulation parameters and process variables on process and product quality. Understanding the variability of both material attributes and process parameters, as well as their overall impact on the unit operations are critical elements for QbD. In a first step, the QbD-methodology is systematically used to
- (1) establish the critical quality attributes representative for the selected dosage forms,
- (2) identify potentially critical input factors that may affect process and product quality and
- (3) risk-rank these factors to define activities for process characterization.
For a blending process the concentration of the active pharmaceutical ingredient, as well as the number of revolutions are related to the blending uniformity. Within the blending simulation approach, several critical points of interest can be effectively addressed by the simulation, e.g. sample locations representing potential areas of poor blending and avoidance of sampling errors. A relationship between low particle number simulated systems and regulatory requirements has been established. By using robust validation strategies, e.g. by the application of PAT, the simulation results can be well correlated to real-life data. A virtual design space is set up, identifying the optimal factor combinations to obtain the best blending output within the given system.
A CFD-simulation approach is used to characterize a spray coating unit operation. Here, the aim is to provide science-based and quantitative understanding of how physicochemical parameters affect the uniformity of the coating layer on a single tablet, passing the spray zone one time at a defined angle. Data then can be used to characterize more complex systems, e.g. tablets passing the spray zone several times at different angles, and knowledge can be leveraged to whole tablet beds. Two potentially critical coating process parameters, i.e. atomizing pressure and inlet temperature, are evaluated and their effect on coating homogeneity is analyzed. By using data provided by the simulation approach, a coating process virtual design space is established. Also here, the optimal combination of input parameters can be identified.
It can be seen that these combined QbD-simulation approaches can well contribute towards an overall pharmaceutical process characterization and highlight several points to consider (e.g. further development of robust validation procedures for models; more efficient simulations that will take into account a higher number of possible input factors) that will be addressed in our future studies.
zurück
