FDA encourages drug manufacturers to apply process analytical technology (PAT) in biomanufacturing processes to provide safe, consistent, and high-quality products for patients. Raman spectroscopy has emerged as a real-time bioreactor monitoring/controlling tool due to its ease of use, fast analysis, and non-destructive nature. A major challenge for the application of Raman spectroscopy is to build robust calibration models (R<sup>2</sup> &gt; 0.9) to accurately predict cell culture parameters such as cell density, metabolites, and multiple amino acids (e.g., L-Lysine, L-Histidine, and L-Arginine). Using data from multiple batches (typically, 10 – 20) can help to obtain accurate calibration models in development space. However, the overall Raman spectrum collection process for 10 – 20 batches usually takes 3 months or longer if only one Raman instrument is available. With the aim to improve the efficiency in developing Raman-based calibration models, we demonstrated that using an autosampler could significantly reduce the number of batches and time required for calibration model development. Autosampler enabled multiple sampling per day without interrupting standard bioreactor operational flows. In a case study, we showed that 4 batches were sufficient in developing satisfying calibration models. At the end, titration studies of viable cell density, titer, and multiple amino acids were conducted to assess the reliability of predictive mathematical relationships.<br/>

Use an autosampler to improve efficiency in developing Raman-based calibration models for mammalian cell cultures

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.