Elucidating molecular properties of kappa-carrageenan as critical material attributes contributing to drug dissolution from pellets with a multivariate approach

Abstract

Multivariate data analysis (MVDA) and artificial neural networks (ANN) are supporting statistical methodologies required for successful development and manufacturing of drug products. To address this purpose, a complex dataset from 49 industrially produced capsules filled with pellets was first analyzed through the development of a multiple linear regression model focused on determining raw material attributes or process parameters with a significant impact on drug dissolution. Based on the model, the following molecular and micrometrics properties of K-carrageenan have been identified as critical material attributes with the highest contribution to drug dissolution: molecular weight and polydispersity index, viscosity, content of potassium ions, wettability, particle size, and density. The process parameters identified to control the drug dissolution behavior of pellets were amount of granulation liquid, torque of dry blend, spheronization parameters, and yields after screening. To further scrutinize the dataset, an ANN model was subsequently built, incorporating 29 batches addressing drug particle size and process parameters such as torque during granulation and spheronization time as critical factors. Finally, this study demonstrates the ability of MVDA and ANN to allow prediction of the key performance drivers influencing the drug dissolution of industrially developed capsules filled with pellets and it highlights their complementary relationship.