Process Raman as a comprehensive solution for downstream buffer workflow

Significance of the topic

Raman spectroscopy as an in-line Process Analytical Technology (PAT) offers rapid, non-destructive and molecule-specific monitoring in aqueous streams without sample preparation. For downstream biopharmaceutical operations such as ultrafiltration/diafiltration (UF/DF), the ability to quantify formulation excipients (e.g., L-histidine, L-arginine, sucrose) and to assess buffer quality in real time can materially reduce reliance on off-line laboratory assays, prevent batch failures, and enable tighter process control and potential automation of critical steps.

Objectives and overview of the study

The work demonstrates the use of the Thermo Scientific MarqMetrix All‑In‑One Process Raman Analyzer to:

• Develop and validate chemometric models to quantify L-histidine, L-arginine, and sucrose in complex matrices relevant to IgG1 monoclonal antibody (mAb) UF/DF processes.
• Demonstrate model transferability across instruments and apply models in-line to a bench-scale UF/DF campaign.
• Showcase the capability of process Raman to detect buffer degradation (sucrose hydrolysis) and provide real-time buffer quality assessment.

Methodology

Key experimental design and data processing steps included:

• Calibration design: Uniform Design (UD) DoE to span concentration space efficiently for mixtures of L-histidine (0–15 mg/mL), L-arginine (0–40 mg/mL) and sucrose (0–200 mg/mL), including protein (IgG1 mAb) spectra in the training set to improve specificity.
• Spectral acquisition: FlowCell and BallProbe sampling optics with laser power 450 mW, integration time 3000 ms, average of 3 spectra (≈18 s total collection per spectrum), acquisition range interpolated to 60–3250 cm−1 with analysis focused on 800–3235 cm−1.
• Preprocessing: relative y‑axis standardization (NIST SRM protocol), normalization using infinity norm in 2900–3230 cm−1 region, Savitzky–Golay (1st derivative, 2nd order polynomial, window = 13), mean centering.
• Chemometrics: Partial Least Squares (PLS) regression with latent variable selection via leave‑one‑out cross‑validation (LOOCV); model validation used independent samples collected on three different instruments to assess transferability.
• UF/DF test conditions: bench‑scale PES membrane, tris buffer pH 7.0 equilibration, feed IgG1 mAb at 10 g/L targeting membrane loading of 500 g/m2; UF feed rate ~300 L/m2/hr with TMP held at 10–15 psi; manual control of diafiltration buffer addition to maintain volume.

Used instrumentation

The experimental platform comprised:

• Thermo Scientific MarqMetrix All‑In‑One Process Raman Analyzer.
• Thermo Scientific MarqMetrix FlowCell sampling optic (in‑line flow measurements).
• MarqMetrix BallProbe sampling optic (for stability/hold measurements).

Main results and discussion

Model performance and in‑line application:

• PLS models achieved high accuracy for all three excipients. Correlation coefficients exceeded 0.99 for calibration and cross‑validation and ~0.995 for prediction; absolute prediction errors at the end of diafiltration were <5% relative to HPLC references (example: L‑histidine 1.0% error; L‑arginine 1.4%; sucrose 3.4%). Reported error metrics included RMSEC ≈0.44, RMSECV ≈0.45 and RMSEP ≈1.13 for one analyte, with comparably low errors for others.
• Models were transferable across three instruments after spectral standardization, supporting practical deployment across process analyzers.
• Applied in a bench‑scale UF/DF campaign, Raman predictions tracked dynamic concentration changes for L‑histidine, L‑arginine and sucrose and captured expected excipient removal behaviour during UF/DF steps.

Buffer stability and quality monitoring:

• During a 15‑day hold of the excipient buffer at room temperature (brief air exposure), Raman predicted a decrease in sucrose from ~86 mg/mL to ~57 mg/mL over 14–15 days.
• Spectral changes showing attenuation of sucrose‑specific bands (e.g., ~835 cm−1 and ~550 cm−1) accompanied by growth of glucose (~525 cm−1) and fructose (~640 cm−1) bands demonstrated sucrose acid hydrolysis into glucose and fructose; HPLC confirmed these species and that histidine and arginine remained intact.
• Concomitant physicochemical changes were recorded: an increase in osmolarity (≈40%) and a pH drop of ~1 unit during the hold period, implicating acidic hydrolysis likely driven by pH change from external factors (e.g., CO2 dissolution, microbial activity, or preparation error).
• Model diagnostics: the Q residual (residual spectral variance not described by the PLS model) increased as glucose/fructose features grew. Reduced Hotelling T2 vs reduced Q plots provided useful control limits to flag buffers failing quality thresholds; spectra after day 5 failed the chosen QC region in this case.

Benefits and practical applications

Process Raman provided multiple concrete advantages:

• Real‑time quantification of key excipients during UF/DF with laboratory‑comparable accuracy, enabling immediate process decisions (e.g., when to stop diafiltration) rather than relying on theoretical diavolume calculations that assume zero retention.
• Ability to detect buffer degradation before use, preventing potential batch failures and reducing time and cost associated with off‑line testing.
• Single‑scan simultaneous measurement of multiple critical process parameters (CPPs)—protein and excipient concentrations and spectral indicators of unexpected species—facilitating tighter control and paving the way for automation and closed‑loop control of downstream operations.

Limitations and considerations

• PLS models built without explicit spectral representations of degradation products (glucose, fructose) can misattribute spectral contributions, leading to overprediction of sucrose when hydrolysis products are present. Model augmentation with relevant degradation species is recommended for robust long‑term monitoring.
• The demonstrated UF/DF runs used manual TMP and feed control; full automation would require integrating Raman outputs into process control loops and validating response strategies.
• Root causes for in‑buffer pH decline were not fully resolved; users must consider aseptic handling, buffer preparation controls, and environmental exposures that can accelerate degradation.

Future trends and potential applications

Expected directions to increase impact and robustness include:

• Expanding calibration libraries to include common degradation products, excipient variants and broader protein matrices to improve model specificity and reduce false alarms.
• Implementing automated feedback control of UF/DF using Raman‑derived real‑time measurements to optimize diafiltration volumes, membrane usage and process time.
• Combining Raman with orthogonal in‑line sensors (pH, conductivity, UV/Vis, turbidity) and digital twins to enable multivariate process control and predictive maintenance.
• Advancing model transfer strategies (standardization, domain adaptation) to simplify deployment across multiple analyzers and plants.
• Regulatory acceptance pathways for PAT‑driven control strategies and data integrity practices to support routine use in GMP environments.

Conclusion

This application study demonstrates that process Raman spectroscopy, implemented with appropriate chemometrics and instrument standardization, can quantify key excipients in real time during UF/DF with errors <5% relative to HPLC and can detect buffer degradation events (sucrose hydrolysis) via spectral diagnostics. These capabilities support improved process control, reduced reliance on off‑line assays, and provide a clear route toward automated downstream processing when integrated with control systems and expanded spectral models.

References

1. Zhang L.; Liang Y.‑Z.; Jiang J.‑H.; Yu R.‑Q.; Fang K.‑T. Uniform Design Applied to Nonlinear Multivariate Calibration by ANN. Analytica Chimica Acta 1998, 370(1), 65–77.
2. Choquette S. J.; Etz E. S.; Hurst W. S.; Blackburn D. H.; Leigh S. D. Relative Intensity Correction of Raman Spectrometers: NIST SRMs 2241–2243 for 785 nm, 532 nm, and 488.5/514.5 nm Excitation. Applied Spectroscopy 2007, 61(2), 117–129.
3. Torres A. P.; Oliveira F. A. R.; Silva C. L. M.; Fortuna S. P. The Influence of pH on the Kinetics of Acid Hydrolysis of Sucrose. Journal of Food Process Engineering 1994, 17(2), 191–208.
4. Wiercigroch E.; Szafraniec E.; Czamara K.; Pacia M. Z.; Majzner K.; Kochan K.; Kaczor A.; Baranska M.; Malek K. Raman and Infrared Spectroscopy of Carbohydrates: A Review. Spectrochimica Acta Part A 2017, 185, 317–335.
5. Kumar S.; Martin E. B.; Morris J. Detection of Process Model Change in PLS Based Performance Monitoring. IFAC Proceedings Volumes 2002, 35(1), 125–130.
6. Agrawal P.; Wilkstein K.; Guinn E.; Mason M.; Serrano Martinez C. I.; Saylae J. A Review of Tangential Flow Filtration: Process Development and Applications in the Pharmaceutical Industry. Organic Process Research & Development 2023, 27(4), 571–591.
7. Nolasco M.; Pleitt K.; Khadka N. Using a Process Raman Analyzer as an In‑Line Tool for Accurate Protein Quantification in Downstream Processes. (Application note).
8. Nolasco M.; Pleitt K.; Khadka N. Raman‑Based Accurate Protein Quantification in a Matrix That Interferes with UV‑Vis Measurement. (Application note).