Facilitating Raman Model Calibration Using Automated Sampling Technologies

Raman spectroscopy is an integral part of a process analytical technology (PAT) strategy to manage the process performance of a bioreactor and ensure the final product meets prespecified quality requirements. The process of Raman calibration typically requires extensive manual sample collection which is time- and resource-intensive and increases the risk of bioreactor contamination.

This page describes the use of an innovative workflow to expedite integration of in-line Raman measurement for a bioreactor application using MAST^® automated sampling and data analysis. This approach accelerates the Raman model-building phase which then enables in-line, real-time monitoring of critical process parameters (CPPs) and critical quality attributes (CQAs) during bioprocessing.

Section Overview

• Experimental Design of the Bioreactor Process, Autosampling, and Raman Spectroscopy Platform
• Integrated ProCellics™ Raman Analyzer and MAST^® Autosampling Solution
• Results and Raman Spectroscopy Model Calibration 
• Benefits of the Combined Raman Analyzer and Automated Sampling Solution
• Conclusion

Experimental Design of the Bioreactor Process, Autosampling, and Raman Spectroscopy Platform

Intensified seed train cell cultures (N-1; n=4 runs) were ran in parallel in four 3 L glass stirred tank bioreactor systems for 11 days. Perfusion operation was enabled by the Cellicon^® Filter which provided cell retention and spent media removal in conjunction with the Cellicon^® Controller. The perfusion rate was adjusted daily to achieve a constant cell-specific perfusion rate (CSPR) of 25 pL.cell^-1.day^-1 and the predicted increase of cell density.

Each bioreactor was connected to the MAST^® Autosampling Solution by aseptically welding the tubing from the sterile bioreactor sample line to the tubing of the autoclaved Sample Pilot. MASTconnect software was programmed and controlled the sampling strategy for the duration of the experiment. Bioreactor samples drawn from each of the four bioreactors and delivered directly to the bioanalyzer where nutrients, metabolites, and osmolality were measured. Viable and total cell density (VCD and TCD) were determined using an automated cell staining and counting analyzer system.

Secreted monoclonal antibody (mAb) was quantified using a High-Performance Liquid Chromatography (HPLC) system with a bind-elute, affinity-based separation using a small-scale protein A column (0.1 mL). Bioreactor samples were drawn on-line every six hours, for four samples per 24-hour period, and 48 total samples from each bioreactor.

The ProCellics™ Raman Analyzer multi-channel unit with four probes was used in this study. Each bioreactor was equipped with one probe, enabling in-line spectral measurements to be taken every 10 minutes in a sequential manner. Bio4C^® PAT Raman Software and Bio4C^® PAT Chemometric Expert Software were used to build the Raman calibration model for mAb titer, glucose, lactate, VCD, and TDC. The models were validated and used to monitor CPPs and CQAs in real time.

Integrated ProCellics™ Raman Analyzer and MAST^® Autosampling Solution

The integrated ProCellics™ Raman Analyzer and MAST^® Autosampling Solution enhanced the Raman model building phase. The traditional off-line process was significantly accelerated with the use of automated sample collection and analytics. The collected data were then input into the Bio4C^® PAT Raman Software, facilitating the implementation of an in-line, real-time Raman sensor in a bioreactor application.

Results and Raman Spectroscopy Model Calibration  

Calibration Raman models that exhibit comparable performance to the traditional approach were developed (see ‘Implementation of Raman Spectroscopy for in-line monitoring of critical process parameters of CHO cell perfusion cultures’ application note). The five calibration models demonstrated a coefficient of correlation (R²) higher than 0.98 and satisfactory performance during cross-validation (Table 1). An independent batch that included 38 sampling points was used as a validation set to evaluate the performance of each of the parameters (VCD, TCD, glucose, lactate, and mAb titer), ensuring robustness and specificity of each model. Evaluation of the root mean square error of prediction (RMSEP) and the relative errors from the validation set provided valuable insights into the capabilities of Raman spectroscopy for monitoring CPPs and CQAs.

Benefits of the Combined Raman Analyzer and Automated Sampling Solution 

The optimized process development method consisting of four bioreactors, each equipped with autosampling and Raman spectroscopy sensor capabilities, delivers several benefits including:

• Accelerated Raman spectroscopy deployment: Raman calibration using the MAST^® Autosampling Solution is 4x less time-consuming compared to traditional approaches; four experiments can run in parallel. Combined with the MAST^® Autosampling Solution, the increased data volume enables rapid construction and validation of the Raman calibration model. Process scaling is expedited with in-line monitoring due to the faster and more reliable model development.
• Enhanced efficiency and consistency: The MAST^® Autosampling Solution enables a more comprehensive mapping of the design space through the acquisition of a larger dataset and captures more complete batch kinetics during the process (Figure 2). With this thorough dataset, a refined understanding of process variations over time is facilitated. Combined with the ProCellics™ Raman Analyzer, robust Raman calibration models can be generated with a minimum number of experiments, encompassing the kinetics design space for each CPP and CQA.
• Ensured reliability and reproducibility: Automation of sample collection and integration of four Raman probes from the same analyzer, enables simultaneous monitoring of four bioreactor batches, offering significant time and labor savings. This approach ensures consistent and reproducible sampling and spectral data collection. Implementing robust Raman models on the ProCellics™ Raman Analyzer is crucial for confident, high-quality monitoring of CPPs and CQAs across the entire process - from development to manufacturing.
• Contamination prevention: The MAST^® Autosampling Solution is connected to the analyzers in a closed and aseptic manner. Combined with implementation of the Raman probe directly in-line with the process, this eliminates the risk of contamination typically associated with manual sampling.
• Autonomous 24/7 operation: The MAST^® Autosampling Solution and ProCellics™ Raman Analyzer can be operated in a continuous manner without the human interaction, significantly enhancing productivity and efficiency.

Conclusion

The innovative combination of the ProCellics™ Raman Analyzer and MAST^® Autosampling Solution creates a highly efficient, reliable, and reproducible approach to spectral data acquisition, sampling, and analysis. By accelerating spectral data acquisition, automating sample collection, and streamlining the analysis process, the platform significantly reduces the time needed for Raman model development and validation. As a result, implementation of Raman spectroscopy from process development to the manufacturing floor is accelerated, playing a pivotal role in creation of a bioprocessing workflow that can compress timelines and streamline delivery of new molecules to the market.