Abstract
[Display omitted]
•Multivariate linear regression model to predict the dissolution profile from NIR.•Fast predictions, independent of process parameters to enable real time release testing.•Model dependent and model independent approaches compared.
A method for predicting dissolution profiles of directly compressed tablets for a fixed sustained release formulation manufactured in a continuous direct compaction (CDC) system is presented. The methodology enables real-time release testing (RTRt). Tablets were made at a target drug concentration of 9% Acetaminophen, containing 90% lactose and 1% Magnesium Stearate, and at a target compression force of 24kN. A model for predicting dissolution profiles was developed using a 34−1 fractional factorial experimental design built around this targeted condition. Four variables were included: API concentration (low, medium, high), blender speed (150rpm, 200rpm, 250rpm), feed frame speed (20rpm, 25rpm, 30rpm), compaction force (8KN, 16KN, 24KN). The tablets thus obtained were scanned at-line in transmission mode using Near IR spectroscopy. The dissolution profiles were described using two approaches, a model-independent “shape and level” method, and a model-dependent approach based on Weibull’s model. Multivariate regression was built between the NIR scores as the predictor variables and the dissolution profile parameters as the response. The model successfully predicted the dissolution profiles of the individual tablets (similarity factor, f2 ∼72) manufactured at the targeted set point. This is a first ever published manuscript addressing RTRt for dissolution prediction in continuous manufacturing, a novel and state of art technique for tablet manufacturing.