Conference Agenda

Monoclonal antibody (mAb) drugs offer advantages such as higher affinity and specificity as compared to conventional drugs for the treatment of critical diseases. Such advantages have led to a rapid growth of the global mAb market, along which new production technologies have been developed including host cell lines such as a high-yield CHO-MK¹ with complex metabolism. In the cultivation step, not only the final product, mAb, but also impurities such as host cell proteins (HCP), DNA, and charge variants are produced. These impurities have a significant impact on quality in the time- and resource-intensive cultivation step, making this step a major factor influencing the overall production cost, time, and quality. Therefore, mathematical models which can reduce experimental burden are useful for the design and operation of this critical step.

In the field of process systems engineering, model-based approaches have been applied to mAb production processes. Several studies have worked on improving mechanistic models based on the understanding of phenomena using data-driven models, e.g., improving lactate model equations with clustering². Some studies have focused on process design by utilizing models, e.g., dynamic optimization for maximizing mAb production while keeping costs low³, and comparison between production processes considering time and costs⁴. However, elements affecting quality described by previous models are limited, which is a significant barrier in the design of mAb production regulated by quality standards.

This work presents a hybrid-model-based approach to CHO-MK cell cultivation process design. Automated cultivation equipment was setup that contains 12 parallel 250 mL bioreactors, and three cycles of fed-batch cultivation were performed by varying agitation speed (700, 1200, and 1400 rpm), dissolved oxygen (20 and 50 %), and glucose feed rate (6, 15, and 20 g L^-1 day^-1). Multiple items including viable cell density, product mAb, metabolites, and impurities were measured as time series data. Based on the experimental data, we first worked on model development using mechanistic model equations from the literature⁵, but the model was unable to reproduce the behavior of lactate and viable cell density. In improving the model, it is difficult to elucidate all the biological phenomena, and increasing the number of estimated parameters is not advisable. Therefore, we developed a hybrid model that maintains the mass balance of the original model while estimating the coefficients using a data-driven model. The developed hybrid model accurately described not only the behavior of viable cell density, lactate, and the final product mAb, but also impurities (HCP, DNA, and charge variants) comprehensively. By utilizing the developed model, we found the condition to maximize mAb while keeping impurities low. In the ongoing work, we are conducting validation experiments for the developed hybrid model.

1. K. Masuda, et al, J. Biosci. Bioeng. (2024), 137, 471-479
2. K. Okamura, et al, Comput. Chem. Eng. (2024), 191, 108822
3. W. Jones & D. Gerogiorgis, Comput. Chem. Eng. (2022), 165, 107855
4. S. Badr, et al, Comput. Chem. Eng. (2021), 153, 107422
5. Z. Xing, et al, Biotechnol. Prog. (2010), 26, 208-219

In cell cultivation, the physiological conditions inside the reactor are critical for achieving the best overall process performance. To reach high titers and productivity, bioprocess engineers try to provide the microorganisms with the best possible environment for the given objective. The dissolved oxygen (DO) level is integral to this, as limitations cause shifts in the metabolic activity of the cultivated organisms or even cell death. This process parameter is commonly manipulated by the stirring speed and the aeration flow rate, where controllers are employed to keep it above a certain threshold. Often simple PID algorithms are deployed for this task, which are then extended with feedback linearization, cascaded control or gain scheduling to tackle the inherent nonlinear nature of bioprocesses (Babuška et al. 2003). Still, when faced with abrupt changes in nutrient addition, the purely reactive nature these systems results in sudden drops of the DO signal and extended periods of oxygen limitation. This problem is encountered in systems where the substrate is added with intermittent bolus shots, such as high-throughput small scale multi-reactor systems paired with a pipetting robot, as described by Kim et al. (2023). These metabolic challenges for the cells can affect their physiological health and further the productivity of the process, which gives the need for a more advanced control scheme.

Model Predictive Control (MPC) emerges as a promising alternative to prevent oxygen limitations using a rather simple model for the non-linear process dynamics and providing it with the known feed trajectory. The resulting oxygen uptake is modeled by combining first principal mass balances and a Monod-type kinetic for the metabolic activity. The oxygen uptake rate is then connected to the physical oxygen transfer rate of the provided air into the liquid phase through the k_LA correlation proposed by Van’t Riet (1979). The parameter estimation is done with lab-scale experiments, where a combination of offline and online measurements is recorded. The MPC algorithm is then tested in-silico with different configurations for the objective function and compared to the performance of a common PID control. The simulated performance of the predictive controller shows that the time of oxygen limitation throughout the process can be minimized by anticipating the needed control action before the next feed pulse is added. These results show that the proposed algorithm can ensure an improved dissolved oxygen level throughout the changing dynamics of the bioprocess, even when challenged with sudden changes in nutrient supply.

Babuška, R., Damen, M.R., Hellinga, C., Maarleveld, H., 2003. Intelligent adaptive control of bioreactors. Journal of Intelligent Manufacturing 14, 255–265. https://doi.org/10.1023/A:1022963716905

Kim, J.W., Krausch, N., Aizpuru, J., Barz, T., Lucia, S., Neubauer, P., Cruz Bournazou, M.N., 2023. Model predictive control and moving horizon estimation for adaptive optimal bolus feeding in high-throughput cultivation of E. coli. Computers & Chemical Engineering 172, 108158. https://doi.org/10.1016/j.compchemeng.2023.108158

Van’t Riet, K., 1979. Review of Measuring Methods and Results in Nonviscous Gas-Liquid Mass Transfer in Stirred Vessels. Ind. Eng. Chem. Proc. Des. Dev. 18, 357–364. https://doi.org/10.1021/i260071a001

This study aims at accelerating the identification of poorly soluble drugs during the phase of drug development, and reducing time and resources needed for experimentation. Typically, the equilibrium solubility of a drug in the gastrointestinal tract is a key parameter for assessing the availability of the drug through the human body. However, the complexity of the extraction and manipulation of human intestinal fluids determines that only few experiments can be carried out in vivo on patients. For this reason, predicting solubility in vitro on Simulated Intestinal Fluids is of paramount importance. To this purpose, pharmacokinetics physiologically based (PBPK) models are utilized. PBPK models describe mathematically the human body by dividing it into a series of compartments, which correspond to different organs or tissues (Stamatopoulos, 2022). However, several phenomena (e.g. the impact of food) are not accounted for, thus leading to inaccurate predictions.

In this study, a novel hybrid modelling approach is proposed, which exploits Gaussian Process and Multi-Linear regression to increase the predictive accuracy and the physical interpretability of the physiological model. The proposed methodology is applied on an Active Pharmaceutical Ingredient whose solubility is measured in fasted and fed conditions (Stamatopoulos et al. 2023). Results demonstrate that the proposed hybrid model outperforms the state-of-the-art literature models, describing both inter- and intra-subject variability of drug solubility in the gastrointestinal tract in a very accurate manner: the determination coefficient in prediction for test sets is = 0.96. The proposed model represents a significant step forward in improving the understanding of the relationship between intestinal components and drug solubility, and in enhancing physiological interpretability.

References

Stamatopoulos, K., Ferrini, P., Nguyen, D., Zhang, Y., Butler, J. M., Hall, J., & Mistry, N. (2023). Integrating In Vitro Biopharmaceutics into Physiologically Based Biopharmaceutic Model (PBBM) to Predict Food Effect of BCS IV Zwitterionic Drug (GSK3640254). Pharmaceutics, 15(2), Article 2. https://doi.org/10.3390/pharmaceutics15020521

Stamatopoulos, K. (2022). Integrating Biopharmaceutics to Predict Oral Absorption Using PBPK Modeling. In Biopharmaceutics (pp. 189–203). https://doi.org/10.1002/9781119678366.ch12