(373a) Use of Model Discrimination Method in Drug Substance Process Development

Drug substance process development often includes several reaction steps to produce the Active Pharmaceutical Ingredient (API), where each step leads to a desired API intermediate. Mathematical models with different structures can be built to represent the proposed reaction mechanisms. It is common for multiple model constructs to equally explain the data gathered at this early stage. At this point it is important to select a model structure among the multiple candidates proposed. This work highlights the application of a model discrimination method to achieve this goal.

The model discrimination method was used to identify a discriminative experimental recipe for a reaction system of interest [1]. The experiment was then conducted to generate data that was qualitatively compared against the model predictions. The model that better explains the trends of the experimental data is deemed to be the best representation of the system among the candidates. To illustrate the use of this method, case studies for different reactions in drug substance process development will be given.

Once a model is discriminated, further investigation can be performed to optimally parameterize the model by following a model based experimental design method. A representative model can help the process development team determine safer and more efficient operating conditions.

Reference

[1] Espie, D. and Macchietto, S. (1989), The optimal design of dynamic experiments. AIChE J., 35: 223–229. doi:10.1002/aic.690350206.