A Model Based Approach to Robustness Analysis and Optimisation of Chromatography Processes

Sammanfattning: One of the major issues in the pharmaceutical industry today is the cost of development and validation of the production of the active pharmaceutical ingredients. The development, optimization and validation of a production plant have been dominated by an extensive experimental approach. The large amount of experimental work and reluctance for changing existing processes has caused the pharmaceutical companies to develop fewer pharmaceuticals and at higher cost today compared to the previous decades. The United States Food and Drug Administration (FDA) has recognised this tendency and realised that they are one part of the problem causing the pharmaceutical companies to be less profitable with fewer new drugs on the market. As a first step to contribute to the modernisation of the pharmaceutical industry the FDA launched a guideline for implementation of Process Analytical Technologies (PAT). A model based methodology provides means to interpret experimental results and formulate them as process knowledge. The ability to predict process behaviour makes it possible to more easily design an operating point that combines process economy and robustness. The impact of process variations can be predicted and measures can be taken to compensate for the impact on the quality of the output. The present work focuses on ion exchange and hydrophobic interaction chromatography and the column behaviour is described by using a kinetic dispersive model, where the steric mass action or langmuir MPM model accounts for the adsorption. This work strives at presenting a model based methodology to efficiently utilise the experimental work normally conducted at the development laboratories to determine a suitable operating point and to determine the robustness of that point. Another contribution is the implementation of robust design where process variations and model error is incorporated into the design to ensure a robust operating point.