A novel sample selection strategy by near-infrared spectroscopy-based high throughput tablet tester for content...

This article proposes a new sample selection strategy to simplify the traditional content uniformity (CU) test in early research and development (R&D) with improved statistical confidence. This strategy originated from the prescreening of a large amount of tablets by a near-infrared spectroscopy (NIRS)-based high-volume tablet tester to the selection of extreme tablets with highest, medium, and lowest content of active pharmaceutical ingredient (API) for further high-performance liquid chromatography (HPLC) test. The NIRS-based high-volume tablet tester was equipped with an internally developed and integrated automated bagging and labeling system, allowing the traceability of every individual tablet by its measured physical and chemical signatures. A qualitative NIR model was used to translate spectral information to a concentration-related metric, that is scores, which allowed the selection of those extreme tablets. This sample selection strategy of extreme tablets was shown to provide equivalent representation of CU in the process compared with the traditional CU test using a large number of random samples. Because it only requires reference tests on three extreme samples per stratified location, the time- and labor-saving nature of this strategy is advantageous for CU test in early R&D. The extreme sampling approach is also shown to outperform random sampling with respect to statistical confidence for representing the process variation. In addition, a chemometric approach, which utilizes only pure component raw materials to develop an NIRS model sensitive to API concentration, is discussed with the advantage that it does not require tablets at multiple API levels. Prospective applications of this sample selection strategy are also addressed.