Feasibility and performance of cross-clone Raman calibration models in CHO cultivation

Area of Science:

• Biotechnology
• Process Analytical Technology (PAT)
• Biopharmaceutical Manufacturing

Background:

• Raman spectroscopy is established for monitoring biopharmaceutical cell cultures.
• Its application in cell line development (CLD) for monitoring culture parameters is less explored.
• The influence of clonal variations on Raman model performance in CLD is not well understood.

Purpose of the Study:

• To investigate the impact of clonal differences on Raman spectroscopy calibration models for predicting titer, glucose, and lactate.
• To assess the accuracy and transferability of these models in the cell line development stage.
• To identify potential biases in Raman models arising from clonal heterogeneity.

Main Methods:

• Development of Raman OPLS models for predicting titer, glucose, and lactate.
• Utilized eleven Chinese Hamster Ovary (CHO) clones with diverse productivity and growth rates.
• Evaluated clone-related biases using clone-wise linear regression on cross-validated predictions.

Main Results:

• Raman models accurately predicted glucose and lactate, unaffected by clonal differences.
• Titer prediction models exhibited significant clone-related bias, linked to productivity variations.
• Model transferability was compromised for titer prediction when applied to cultures outside the training data's productivity range.

Conclusions:

• Raman spectroscopy is a feasible tool for accurate glucose and lactate monitoring during cell line development.
• Accurate titer prediction using Raman spectroscopy necessitates accounting for specific clonal characteristics.
• Clonal productivity significantly influences the accuracy and transferability of Raman-based titer prediction models.