(107c) Leveraging HPLC Data in a QbD Approach, for Right-First Time Scale-up of API Reaction Steps

HPLC data that are routinely generated can be leveraged to deliver much additional value; when interpreted quantitatively, they can be used to develop mechanistic models.  These models in turn can be used to probe and question the data in a way that leads to increased process understanding. There is also the opportunity to run experiments that are designed to produce impurity so that the mechanism of impurity formation can be understood.  The above are core elements in developing a process that works right first time on scale-up, achieving Quality by Design.

In more detail:

Experimental data are the basis for adjusting rate parameters in kinetic models. The most powerful data sets are analytical profiles, as these data are able to show species levels lower than 1%, particularly important in the Pharmaceutical Industry. At the same time these are one of the most difficult to handle, as chromatographic detection is in the case of the most common UV-detectors linked to molecular absorbance, and this can vary by orders of magnitude.

The talk will present an approach to convert analytical data to absolute values by use of Relative Response Factors and a link to the mass balance implied in a kinetic model. The data and model may not agree and in this case either the data or the model may be wrong. 

By changing one or the other, we can ultimately reach data consistent with the model's mass balance, learning much in the process about our analytical method and our reaction scheme; this consistency is a requirement to get a good match in fitting with kinetic models,. The other factor is the rate law assumption, which we also address in this talk.

In an enhancement of this approach, a method is proposed which allows the calculation of the Relative Response Factors under the conditions of overlapping mass balances, which is very often the case.