After the safety and efficacy of a biopharmaceutical, the efficiency of bioprocessing it might be the most crucial aspect. As Aleksandr Pogodaev, PhD—principal scientist in pharmacometrics, Novartis Technical Research & Development, Basel, Switzerland—and his colleagues reported: “Improving bioprocess efficiency is important to reduce the current costs of biologics on the market, bring them faster to the market, and to improve the environmental footprint.” Nonetheless, these Novartis scientists noted that the pre-stages of bioprocessing have largely been neglected in terms of improving efficiency.

Instead of using a batch-fed process, a perfusion process can be more efficient. As Pogodaev and his colleagues noted: “Over the past decade, the use of perfusion during cell cultivation of the pre‐stages (N‐1, N‐2, corresponding accordingly to 1 or 2 stages before the main stage) has gained popularity in the industry.” These pre-stages of bioprocessing aim to produce the highest cell density in a culture.

So, the Novartis team explored the use of modeling perfusion profiles in the pre-stages of bioprocessing a monoclonal antibody (mAb). The scientists used process data to develop a probabilistic model of perfusion. For example, first-order ordinary differential equations were used to model the bioprocess dynamics of the system. Plus, the scientists explained that the use of a probabilistic model “considers uncertainty in parameter estimation, biological noise, and measurement error.” The resulting trained model was validated for simulating various pre-stage conditions.

The scientists from Novartis tested this model on the pre-stages of bioprocessing mAbs from different cell lines. Based on these simulations, they pointed out that “the specific perfusion profile is critical.” The results also revealed various ways to improve the efficiency of the pre-stages. For instance, the culture medium consumed during the pre-stages could be reduced by 25–45%. In addition, modeling can reduce the number of experiments required to develop a process.

With this modeling approach, the Novartis team concluded: “The experimental workload can be reduced by up to 30–70% during process development and process characterization stages.”

Previous articleDigital Biopharma Facing Data Science Skills Gap
Next articleIncentives for Embracing the Industrial Internet of Things