Raman Spectroscopy for Characterizing Monoclonal Antibody Reduction: A Process Analytical Technology Approach for Antibody-Drug Conjugation Process Development.

BACKGROUND

Controlled disulfide bond reduction is critical in antibody-drug conjugate (ADC) manufacturing, enabling site-specific conjugation and desired drug-to-antibody ratios. Current characterization relies on time-consuming offline capillary electrophoresis, creating bottlenecks in process development where multiple conditions must be rapidly screened. This study evaluates Raman spectroscopy as a process analytical technology (PAT) to elucidate reduction kinetics and accelerate optimization for ADC development.

METHODS

A Design of Experiments approach investigated TCEP-mediated reduction of an IgG1 antibody under varying TCEP/mAb ratios (5-15) and pH conditions (5.5-7.5). Raman spectra were collected throughout reduction reactions. Principal component analysis (PCA) characterized reduction kinetics, while partial least squares (PLS) regression quantified fragment formation against non-reduced capillary electrophoresis sodium dodecyl sulfate (nrCE-SDS) measurements.

RESULTS

PCA effectively captured reduction kinetics, with PC1 trajectories correlating strongly with heavy chain (HC) formation measured by nrCE-SDS. The analysis revealed pH-dependent TCEP saturation effects: higher pH showed convergent kinetics at elevated ratios while lower pH maintained ratio-dependent rates. PLS models successfully predicted HC formation, demonstrating potential for endpoint detection despite the interchain disulfide bonds representing only ~ 0.2% of total protein mass.

CONCLUSIONS

Raman spectroscopy with chemometric analysis provides valuable insights for ADC process development. PCA enables rapid screening of reduction conditions without nrCE-SDS confirmation, reducing analytical burden during early development. Unlike offline nrCE-SDS, Raman offers potential for real-time online PAT implementation. Future work focusing on narrower operating ranges could enhance model performance for manufacturing applications.