Oxygen-based soft-sensing of glucose demand in Chinese hamster ovary fed-batch cultures across scales

Glucose is a critical nutrient in mammalian cell culture processes, influencing cellular metabolism, productivity, and product quality. In industrial Chinese hamster ovary (CHO) fed-batch processes, glucose concentrations are commonly controlled using manual offline measurements and operator-based feed adjustments. Although online Process Analytical Technology (PAT) solutions and oxygen uptake rate (OUR)-based feeding approaches have previously been investigated, their implementation in existing Good Manufacturing Practice (GMP) manufacturing environments remains challenging due to instrumentation, validation, and process integration requirements. In this study, the relationship between cumulative oxygen consumption and glucose utilization was investigated across 58 CHOK1SV GS-KO® fed-batch cultures performed in stirred glass and single-use bioreactors ranging from 10 L to 2 kL scale. Building on previously reported oxygen-based glucose estimation concepts, a lactate-adjusted framework was developed to account for both lactate production and re-consumption during the fed-batch process. Oxygen transfer rates were calculated from standard online process parameters and integrated over time to determine cumulative oxygen consumption, which was subsequently correlated with cumulative glucose utilization. The results demonstrated robust linear correlations between cumulative oxygen consumption and lactate-adjusted cumulative glucose utilization across different projects, cell lines, and process scales (R2 ≥ 0.984). Incorporation of lactate metabolism improved the robustness of the oxygen-glucose relationship, particularly during metabolic transitions associated with lactate re-consumption. The predictive capability of the framework was further evaluated in three 200 L fed-batch processes by prospectively estimating glucose consumption over sequential 24 h intervals using cumulative oxygen consumption and routine offline metabolite measurements. Overall, the study demonstrates that cumulative oxygen consumption provides a robust and scalable basis for oxygen-based soft-sensing of glucose demand in industrial CHO fed-batch cultures. The presented framework enables predictive glucose demand estimation using standard bioreactor process data without requiring additional glucose-specific PAT hardware and may support future implementation of automated feeding concepts in mammalian cell culture processes. KEY POINTS: • Cumulative oxygen consumption correlates linearly with glucose utilization. • Lactate-adjusted O₂-glucose correlations are consistent across bioreactor scales. • Glucose control can be automated using existing bioreactor process data.