Purpose: Intelligent automation of pharmaceutical manufacturing suites offers the opportunity to improve the drug and formulation discovery process. The poster will highlight how a high-speed in-line inspection system can greatly enhance the potential of a completely automated development process of a pharmaceutical oral solid dosage (OSD) form from powder to granules to finished tablets. Methods: Pharmaceutical granules containing micronized Paracetamol as active pharmaceutical ingredient (API) were produced on a Xelum R&D granulation suite (Syntegon, Germany). To generate granules with varying API content, powder blends containing 7.5-9-10-11-12.5% Paracetamol (target values) were prepared, by adapting the amounts of Lactose and MCC, keeping their ratio constant. Moisture and density/particle size were varied by modifying the end point product temperature of drying and the spray air pressure. In total 13 experiments were conducted in duplicate, each in 5 kg batches (26 batches in total). Subsequently the granules were blended with 1% of magnesium stearate and compressed on a rotary tablet press equipped with a feeder for automatic process development (TPR 200, Syntegon, Germany). Besides typical compression parameters, the feeder configuration was automatically varied, while the fill depth was kept constant at 8.0 mm. The tablets were round (diameter: 10 mm), convex, with a debossed marking on one side. The target thickness was 4.2mm. In total 265 different experiments were generated. The products manufactured during these batches had a Target API content from 7.5% to 12.5% in 5 steps, a granule moisture from 1.11% to 1.85%, a granule particle size from 234 µm to 333 µm, a tablet hardness from 46 N to 251 N, a tablet thickness from 3.92 mm to 5.08 mm, a tablet mass from 290 mg to 407 mg, a pre compression force from 0.7 kN to 20 kN and a compression force from 3.5 kN to 25 kN. The tablets were inspected on a CU-120 machine (Pharma Technology, Nivelles, Belgium), capable of 100% testing and sorting of tablets on their thickness via laser scanning, their API fraction via integrated multipoint near-infrared (NIR) spectroscopy, and their mass via 3D microwave resonance technology (3D MRT). All tablets from the 255 batches (over 90000 tablets in total) were nondestructively inspected on the CU-120 machine on their thickness, mass, and API content, with at least one sampling per inspected batch. The thickness of the tablets was measured in line using data provided by laser sensors and interpreted using a proprietary algorithm. Results: The API fraction of every tablet was consecutively predicted via the application of the NIR model on the run, on spectra collected in-line. The accuracy expected for the content predictions is reported on the chart below (right-hand side). The error margin on the label claim (LC, target value) was about 2.38%. The mass of each tablet was finally measured with the microwave sensor. To assess the accuracy and the precision of these measurements, the predicted mass of each sampled tablet was compared to its reference mass (returned by the embedded 4-parameter IPC tester). Conclusion: High-speed in-line inspection can be a key addition for DoE-based and automated development of OSD forms. It provided here fast and valuable data about the produced tablets (data from 90000 tablets collected in less than 3 days), that could further be correlated with the critical process parameters of every single unit operation involved in the manufacturing process. This work has been performed using a machine capable of in-line inspection at production scale, batch recoveries and small batch inspection as well, at a very high inspection rate. The poster will contain the experimental setup, Design of Experiments and the modelling data. The results and obtained accuracies will be discussed.
