Single-pass tangential flow ultrafiltration (SPTFF) is crucial for the continuous manufacturing of monoclonal antibodies (mAbs), but is challenged by variations in mass transfer across the SPTFF membrane, automation and scheduling of operations, and real-time monitoring and control to identify and correct process deviations. A real-time, automated monitoring and control strategy developed by researchers at the Indian Institute of Technology Delhi (IITD) strives to resolve these issues.
This model-based control strategy, detailed in a recent paper in the American Association of Pharmaceutical Scientists Journal, “acts as a digital twin for the SPTFF step,” says Anurag S. Rathore, PhD, coordinator of the DBT Center of Excellence for Biopharmaceutical Technology, IITD. “It includes spectroscopic monitoring sensors and automated control elements that can compensate for feed material variability and process deviations that may arise over long, continuous campaigns.”
This approach can be used two ways for downstream mAb purification. One “independently controls the final formulation step by adjusting the flow rate to achieve fixed retentate concentration targets, and relies on a surge tank prior to the SPTFF step to facilitate independent control of the SPTFF feed flow rate as it is decoupled from the preceding operation,” Rathore says. It can also be used for in-line concentration of intermediate process flows between two purification steps, to enhance productivity.
In experiments, the model determined the operating conditions needed to achieve the required retentate concentrations despite feed material variability, and allowed SPTFF to occur across multiple cycles of continuous processing in a controlled, automated manner. Retentate concentrations varied from the target concentration by no more than 10% in each of the 15 runs performed.
Importantly, this method supports the FDA’s process analytical technology and quality by design guidances, he says, “building flexibility and robustness into the process without the need for frequent human intervention.”
In future iterations, the researchers plan to address membrane fouling and the ability to predict reductions in normalized water permeability through the SPTFF membranes so they can be cleaned at the optimal times. “We also may explore combinations of spectroscopic sensors, including Raman and Fourier-transform infrared spectroscopy or fluorescence sensors, to capture more detailed information about the mAb and the process trajectory through multivariate analytics,” Rathore says.
Eventually, he hopes to “integrate digital twins and model-based controls for all the different unit operations in the downstream train to handle process deviations at any time or location during continuous operation.”
