Effects of Operating Mode and Flow Conditions on Performance of Virus Removal Filters for Therapeutic Antibody Production
Restricted (Penn State Only)
- Author:
- Peles, Joshua
- Graduate Program:
- Chemical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 09, 2024
- Committee Members:
- Robert Rioux, Professor in Charge/Director of Graduate Studies
Andrew Zydney, Chair & Dissertation Advisor
Hee Jeung Oh, Major Field Member
Darrell Velegol, Major Field Member
William Hancock, Outside Unit & Field Member - Keywords:
- Virus Filtration
Therapeutic Antibody
Protein Fouling
Virus Removal
Virus Breakthrough
Downstream Bioprocessing
Continuous Bioprocessing
Process Disruption
Flow Interruption - Abstract:
- There is growing interest in the implementation of continuous processing technologies for manufacture of biotherapeutic products and antibody-related biologics. However, this anticipation for the potential benefits associated with this transition must be balanced by an attention to uncertainties regarding the operation and performance of many critical steps within the downstream bioprocess during continuous manufacturing. Virus removal filtration is one such step that needs assessed; it is currently operated at high and constant transmembrane pressures (TMPs) under traditional batch processing, but it will need to instead be performed at low and constant permeate fluxes under future continuous processing. The objective of this dissertation study was to investigate the effects of operating mode (constant TMP vs constant flux) and flow conditions (high flow vs low flow) on the overall filtration performance and underlying physical behaviors with regard to protein fouling and virus breakthrough with the Viresolve® Pro virus filter, which is one of the most commonly used viral filters in commercial bioprocessing. Filtrate/permeate flux data obtained during constant-TMP filtration experiments with human serum Immunoglobulin G (hIgG) as a model protein foulant at higher TMPs (TMP > 5 psi) were well-described using a combined fouling model with sequential complete pore blockage and cake filtration. hIgG fouling decreased with increasing TMP at initial stages of the filtration and then increased with increasing TMP at later stages. Filtrate flux data at lower TMPs (TMP < 5 psi) were instead described using a combined model with sequential intermediate pore blockage and cake filtration. In both cases, protein fouling was attributed to hIgG aggregates (trimers and higher-order oligomers) that were present at low concentrations and were larger in size than the smallest pores in the Viresolve® Pro filter. The difference in the pore blockage mechanisms at lower and higher TMPs was due to the effects of foulant diffusion at low flux as described by changes in the Péclet number for the foulant aggregates. The rate of the pore blockage decreased with increasing TMP, reflecting the greater drag forces that help push the protein aggregates through the pores. In contrast, the growth rate for the protein cake resistance increased with increasing TMP due to the compressibility of the hIgG deposit, leading to a maximum in the rate of fouling at intermediate TMP around 5 psi. The data from both TMP regimes were described very well using simple functional relations for the values of the model parameters as functions of the operating TMP, providing a global fouling model that is in good agreement with the filtration performance for operating TMPs from 0.2 to 60 psi using only three fitted model parameters. Results from constant-flux filtration experiments were also well-described using the combined sequential pore blockage and cake filtration model using the same model parameters that had been determined under constant-TMP operation. The rate of fouling at lower fluxes (J ≤ 10 LMH) was determined primarily by the rate of the cake’s growth while the rate of fouling at higher fluxes (J ≥ 50 LMH) was determined primarily by the rate of the pore blockage, where the highest rate of fouling occurring at intermediate flux (J = 20-30 LMH). This new model form is able to provide accurate descriptions of filtration performance for operating permeate fluxes of 5-100 LMH, providing a framework that can be used to design and optimize the performance of virus filtration processes for continuous biomanufacturing operations. Virus removal/retention (LRV) data were obtained during extensive constant-flux and limited constant-TMP filtration experiments using ΦX-174 bacteriophage as a model virus particle. The retention data were described very well using an internal virus polarization model, where phage retention decreased with decreasing flux due to the greater mobility of the virus particles within the matrix-like structure of the filter (the “reservoir-zone”). Process disruption experiments with planned flow interruptions showed a significant increase in virus transmission immediately after the flow release; these data were also well-described using the modified form of the prior internal polarization model, with the increased transmission due to the release of phage that were previously captured during the filtration. The fraction of captured particles that were released and their rate of re-capture were both relatively insensitive to many of the process conditions as long as the characteristic virus diffusion distance during the disruption was greater than a threshold of several pore diameters. Phage challenge experiments performed in the presence of hIgG showed that protein fouling had relatively little effect on the virus retention behavior by the Viresolve® Pro beyond the changes in membrane permeability caused by the fouling. The results from this dissertation study provide both fundamental physical insights and deeper practical understanding of the underlying behaviors that govern the overall performance during virus removal filtration, including the manner(s) in which the performance changes with operating mode and flow conditions. In addition, the models that were developed to describe effects of TMP on protein fouling behavior and effects of permeate flux on virus retention behavior provide a framework that can be used to design and optimize virus removal filtration steps as part of continuous downstream processing operations.