Poor solubility of active pharmaceutical ingredients represents an important challenge for pharmaceutical industry in the development of new original and generic drugs. As a way to improve bioavailability of poorly water-soluble active ingredients, solid dispersions are increasingly being utilized. They are defined as dispersions of one or more active ingredients in an inert carrier in a solid state and can be prepared using a wide variety of methods, with solvent-based methods being most commonly employed. This group of methods also include fluidized bed granulation, which was used in our research work. During the process, we sprayed a solution of the active ingredient and polymer onto inert carrier particles, resulting in the formation of granules. We monitored the process along the production line using Raman spectroscopy and performed further analyses in the laboratory using near infrared spectroscopy. Both techniques are most commonly used process analytical technologies for monitoring pharmaceutical manufacturing processes.
Using spectroscopic data from the process analytical technology analysers, we employed multivariate analysis and partial least squares regression to create models for predicting the content and proportion of crystalline active ingredient in granulate samples. We compared the values predicted by the models with the actual values, determined by reference methods. In addition, we verified the comparability of the experiments using the principal component analysis method. We experimentally evaluated the developed models and found acceptable prediction errors and a good fit between the predicted and actual values in all of them. This demonstrated that based on near infrared and Raman spectroscopic measurements, we can create models that can be used in the solid dispersion production processes for monitoring the content of pharmaceutical active ingredient and detecting the potential occurrence of its crystalline forms.
|
