The Use of Raman Spectroscopy for Real-time In situ Monitoring of Multiple Biochemicals During Microbial Process Development

The feasibility study was successful at demonstrating the capability of measuring quantitatively, the real-time concentrations of key biochemicals during a Pichia pastoris fermentation process via the use of in situ Raman spectroscopy. From a process development perspective, utilizing process analytical technology (PAT), such as Raman spectroscopy, to monitor the changes of key biochemicals over time, within the reactor, increases: (i) process knowledge, (ii) provides real-time data to evaluate the process “state of control” to make informed process decisions, and (iii) utilize the predicted biochemical concentrations as automation inputs to operate under normal manufacturing conditions.  This study investigated how changes to the fermentation process – including changes to the cell line, media components and pH set-point – impacted the quantitative model predictions of key biochemicals.  In total, Raman spectra collected from fourteen different fermentation conditions were used for both modeling and validation.  The multivariate modeling approach entailed selecting regions of the spectra sensitive to specific biochemistries as-well-as taking into consideration regions that displayed matrix interactions with other molecules that could affect predicted biochemical concentrations.  Calibration models were developed using partial least squares (PLS) regression employing empirical data (i.e. offline assay concentration values acquired at various time points throughout each fermentation batch) of methanol, glycerol and sorbitol biochemistries.  To confirm the sensitivity of spectral regions, as well as to expand the concentration range and refine the quantitative models, spiking events were also performed in several batches.