Quality Control of fermentation broths

Importance of the Topic

Monitoring quality parameters in fermentation broths is critical for ensuring consistent production of small molecules and protein-based APIs in the pharmaceutical industry. Real-time or near real-time analysis reduces process variability, minimizes batch failures, and accelerates time-to-market. Traditional analytical methods are time-consuming and resource-intensive, creating bottlenecks in modern bioprocessing workflows.

Objectives and Study Overview

• Develop a rapid, multiparameter NIRS approach for key quality attributes in fermentation broths
• Compare predictive performance against standard laboratory techniques (pH meter, UV-Vis, HPLC, PCR)
• Demonstrate feasibility of implementing NIR spectroscopy for routine in-process monitoring

Methodology and Instrumentation

All measurements were carried out on a Metrohm DS2500 Solid Analyzer operating in reflectance mode, requiring no sample pre-treatment due to the dark coloration of fermentation broths. The NIRS transflection vessel and a mini sample cup holder were used for liquid samples. Data acquisition and model development were performed with Vision Air Complete software.

• DS2500 Solid Analyzer (400–2500 nm spectral range)
• Optically flat transflection vessel for liquids
• Mini sample cup holder for routine series measurement
• Vision Air Complete for data management and chemometric modeling

Main Results and Discussion

Prediction models were generated for pH, bacterial concentration, glucose, reducing sugars, and potency. Figures of merit indicate:

• pH: R2 ≈ 0.65, validation error 0.10 pH units
• Bacterial content: R2 ≈ 0.71, validation error ~5 CFU/mL
• Glucose: R2 ≈ 0.92, validation error <1 %
• Reducing sugars: R2 ≈ 0.99, validation error ~1.24 %
• Potency: R2 > 0.90 for both UV-Vis and HPLC + PCR assays, validation errors ~1,168–2,089 u/mL

The strongest correlations were observed for reducing sugars and glucose, underscoring NIRS sensitivity to carbohydrate moieties. Moderate performance for pH and cell density suggests further model refinement or spectral preprocessing could enhance accuracy.

Benefits and Practical Applications

• Analysis time under one minute per sample, compared to minutes to hours for conventional methods
• Reduced consumable costs and elimination of chemical reagents
• Non-destructive measurement preserving sample integrity
• Suitable for harsh production environments due to robust instrument design

Future Trends and Potential Applications

• Implementation as part of process analytical technology (PAT) for continuous monitoring
• Expansion of chemometric models to additional bioprocess parameters such as biomass, metabolites, and product titer
• Integration with in-line probes for fully automated, real-time control
• Advances in machine learning algorithms to further improve prediction accuracy

Conclusion

NIR spectroscopy on the DS2500 platform offers a rapid and cost-effective route to monitor multiple critical quality attributes of fermentation broths. The technique demonstrates sufficient accuracy for routine quality control and has the potential to streamline bioprocess workflows while reducing reliance on traditional laboratory assays.

Reference

• Metrohm AG Application Note AN-NIR-099: Quality Control of Fermentation Broths