Advancing Raman model calibration for perfusion bioprocesses using spiked harvest libraries

Area of Science:

• Biopharmaceutical Process Monitoring
• Spectroscopic Analysis
• Chemometrics

Background:

• Raman spectroscopy is increasingly used for real-time monitoring in biopharmaceutical processes.
• Challenges include model calibration due to limited analyte variability and component cross-correlation.

Purpose of the Study:

• To develop a systematic and efficient Raman calibration workflow for perfusion processes.
• To enable highly specific and rapid model calibration for bioprocess monitoring.

Main Methods:

• Established harvest libraries from multiple CHO cell culture bioreactors.
• Performed offline model calibration using a flow cell with spiked process harvest samples.
• Included sugars like glucose, raffinose, galactose, mannose, and fructose for calibration.

Main Results:

• Raman spectroscopy successfully distinguished and quantified similar sugars in perfusion harvest.
• Developed a robust calibration model for simultaneous glucose and raffinose monitoring (RMSEP: 0.32 g/L and 0.17 g/L, respectively).
• Verified the model's performance during a perfusion process.

Conclusions:

• The novel offline calibration workflow enables Raman peak decoupling and reduces calibration time from months to days.
• This approach is applicable to other critical analytes such as lactate, ammonia, amino acids, and product titer.
• Enhances the utility of Raman spectroscopy for biopharmaceutical process control.