Raman spectroscopic based chemometric models to support a dynamic capacitance based cell culture feeding strategy

Area of Science:

• Biotechnology
• Chemical Engineering
• Process Analytical Technology (PAT)

Background:

• Process Analytical Technology (PAT) tools are increasingly integrated for advanced cell culture monitoring.
• In-line probes offer real-time data crucial for optimizing biopharmaceutical manufacturing.

Purpose of the Study:

• To evaluate the combined use of capacitance and Raman spectroscopy for cell culture monitoring.
• To develop and compare process models using different data analysis strategies for Raman spectroscopy.

Main Methods:

• Utilized in-line capacitance probes for dynamic feeding strategies and Raman spectroscopy for real-time monitoring.
• Collected data from eight 15,000 L scale cell culture batches.
• Developed Partial Least Squares (PLS) models for Raman spectroscopic data, comparing full dataset analysis with data split by capacitance feeding strategy.

Main Results:

• The PLS model using a split dataset based on capacitance feeding strategy yielded a lower RMSEP (1.40 pf/cm) for capacitance compared to the full dataset model (1.54 pf/cm).
• The second Raman modeling method showed results within expected process limits for capacitance, with only a 0.01% difference in total nutrient feed compared to the capacitance probe.
• Raman spectroscopy successfully modeled additional critical process parameters including viable cell density (VCD), viability, average cell diameter, and viable cell volume (VCV).

Conclusions:

• Tandem application of capacitance and Raman spectroscopy provides robust real-time monitoring and control in large-scale cell cultures.
• Splitting Raman spectroscopic data based on capacitance feeding strategy improves model accuracy for predicting capacitance.
• This integrated PAT approach enables precise control over nutrient feeding and accurate prediction of key cell culture attributes.