UF/DF Scale-Down Modeling with Alternative Screen Channel Cassettes
What should be done if the design space established during process characterization is either broader or narrower than the operating parameter ranges used for manufacturing Phase III clinical trial batches?
Ⅰ. Problem Analysis
Consideration: The question concerns the representativeness of a scale-down model for an ultrafiltration/diafiltration (UF/DF) process. Since a C-screen channel ultrafiltration cassette with the required pore size/MWCO is not available, an A-screen channel ultrafiltration cassette is planned to be used as a substitute. The key question is whether the A-screen cassette can serve as a representative scale-down model.
To answer this question, we first need to understand whether there are differences between ultrafiltration cassettes with different screen channel designs and what impact these differences may have on the UF/DF process.
1. Selection of Different Screen Channel Types in Ultrafiltration Cassettes
Regarding the knowledge of different screen channel types in ultrafiltration cassettes, a detailed description is provided in (Document 1: Ultrafiltration/Diafiltration Process Development of High Concentration/Viscosity Applications):
We currently offer four different screens: (A, C, D and V) in the Pellicon® cassette format range.
A screen: A tight screen that is used for dilute protein solution or low viscosity solutions. The tight weave is not suitable for highly concentrated protein solutions as the pressure drop is unacceptably high at greater viscosities. However, this screen typically provides superior flux performance.
C screen: A coarse screen that is used for product streams with viscosity up to 15 cP. The C screen provides good flux performance but the pressure drop is above the limit of usage when the viscosity increases.
V screen: A suspended C screen. The V screen gives lower pressure drops than the C screen, but the flux performance is much lower compared to C screen.
D screen: This new screen has a coarser mesh and altered weave compared to the standard C screen. It was designed to offer lower pressure drop without the large flux performance penalty found when using a suspended screen technology.
Based on the above description of different screen types and publicly available information, the characteristics of different screen channels can be summarized as follows:
2. Impact of Different Screen Channel Types on UF/DF Processes
Based on the descriptions above, it can be reasonably inferred that different types of screen channels directly affect UF/DF process performance:
Fluid dynamics and shear force:
Because the screen (mesh) structure determines the flow pattern of the fluid across the membrane surface, as well as the degree of turbulence and the resulting shear forces generated.
Mass transfer efficiency:
The type of screen affects concentration polarization and the control of membrane fouling.
Process performance:
Such as final flux, retention rate, and product quality (e.g., shear-induced aggregation).
(Reference 2: "Effect of channel-induced shear on biologics during ultrafiltration diafiltration (UF/DF), 2016") demonstrates the impact of shear forces induced by different screen channel designs on the quality of biopharmaceutical products such as monoclonal antibodies and fusion proteins.
In this study, under standardized production conditions of fixed feed flux (300 LMH) and transmembrane pressure (1.4 bar), membrane modules from three manufacturers-MilliporeSigma, Pall, and NovaSep-were evaluated, including medium screen, light-suspension screen, and open-channel configurations. Four representative biologics, including shear-sensitive IgG4, shear-tolerant IgG1, and fusion proteins, were used as test samples. Analytical methods such as SEC, MFI, DLS, and Vmax filterability tests were applied to systematically assess the hydraulic performance of different channel designs and their impact on product quality.
The experimental results show that the channel (screen) structure directly determines shear intensity and mass transfer efficiency.The medium screen channel exhibited the highest mass transfer coefficient; however, both wall shear stress and total shear stress were significantly elevated. In contrast, the open-channel (screenless) configuration generated the lowest shear stress, but suffered from extremely poor mass transfer performance, requiring excessive recirculation cycles and resulting in low process efficiency.The light suspension screen channel (Pellicon D) demonstrated the best overall performance. Its mass transfer coefficient was only 22% lower than that of the medium screen, yet 260% higher than that of the open-channel design. Meanwhile, its total shear stress was the lowest among all configurations, at only 1120 Pa. In terms of product quality, shear-sensitive proteins processed using the medium screen showed a 78%–186% increase in soluble aggregates, along with a significant rise in sub-visible particles. The Vmax value for 0.2 μm sterilizing filtration was very low, indicating severe membrane fouling and poor filterability. By contrast, samples processed with the light suspension screen channel showed only a 0%–25% increase in aggregation, significantly reduced particle levels, and a 3–18× improvement in filtration performance. High-concentration concentration studies further confirmed that the light suspension screen channel maintained stable transmembrane pressure and could easily achieve concentrations above 200 g/L. In comparison, the medium screen channel exhibited a sharp increase in pressure at high concentrations, making it difficult to reach target concentrations and resulting in degraded product quality.
(Reference 3: "Ultrafiltration behavior of recombinant adeno-associated viral vectors used in gene therapy", 2021) compared the impact of C-screen and D-screen channel designs on the UF/DF process of AAV vectors.
The results showed that when the D-screen was used, the critical transmembrane pressure (TMP) was lower, and the average flux was reduced by 14%.
Fig. 2C presents flux versus TMP data using the 30 kDa membrane with the "D" screen. The D screen or V screen would not be suitable for ultrafiltration of low viscosity feed streams in this work. Past work by our group have used these screened channels for high concentration monoclonal antibodies and Fc-fusion proteins [13,33]. For AAV2 with the "D" screen device, the critical TMP was reached at lower TMP than that with the "C" screen device shown in Fig. 2A. Consistent with theoretical expectations, the flux with the "D" screen device was on average 14% lower at feed flow rates of Q = 350 L/h/m2 and Q = 500 L/h/m2 (p < 0.05) compared to the "C" screen device, but there was no difference between the "C" screen (RC2A) and the "D" screen devices (RC2B) at a feed flow rate of Q = 120 L/h/m2 (p > 0.05) (Fig. 2C).
ⅡImplementation recommendations:
It can therefore be concluded that screen type is a critical factor affecting both process performance and product quality, and must be properly considered in scale-down models. Accordingly, in response to the inquiry, if an identical C-screen membrane cassette is not available, it is not appropriate to simply perform a direct substitution based on "same pore size but different screen type," such as replacing it with an A-screen membrane cassette.
Direct substitution (i.e., using the same operating parameters) is generally not representative, especially in the following cases:
· When the product is shear-sensitive (e.g., monoclonal antibodies, fusion proteins, viral vectors).
· When accurate prediction of large-scale process performance is required, including flux, filtration time, and fouling behavior.
· When process characterization studies are conducted to define the design space of critical process parameters (CPPs), such as transmembrane pressure (TMP) and crossflow velocity.
Then are there any other suitable alternative options or implementation measures? After discussion, the Xingchen team recommends the following approach:
1．Identify functionally equivalent alternative membrane cassettes
This involves identifying small-scale membrane cassettes from other suppliers that exhibit similar hydrodynamic characteristics, particularly shear behavior.
It is recommended to consult ultrafiltration membrane cassette suppliers and request laboratory-scale products that are designed based on similar principles and exhibit flow dynamics comparable to the C-screen channel cassette used in production.
Key parameters of candidate cassettes should be compared with the target C-screen cassette, including channel height (or hydraulic diameter), screen structure (e.g., screen type), membrane material, and pore size, as these factors determine the shear forces generated in ultrafiltration cassettes.
2.Use an A-screen cassette and compensate for hydrodynamic differences by adjusting operating parameters
In theory, the A-screen configuration typically has a tighter flow channel than the C-screen, which may result in higher shear stress. Therefore, the shear difference between the two systems under the same feed flow rate should be estimated through calculation or computational fluid dynamics (CFD) simulation.
The feed flow rate (Q) of the small-scale A-screen system should then be adjusted so that the shear stress generated at the membrane surface matches the shear stress produced by the large-scale C-screen system under the target operating flow conditions.
Under the adjusted shear conditions, conduct small-scale experiments to measure key performance indicators such as flux and retention rate. These results should then be compared with historical data or parallel experimental data from the large-scale C-screen system to validate its predictive capability.
3.Focus on specific aspects and establish a reduced representativeness scale-down model
If the focus of process characterization is not on shear-sensitive products, or if the primary objective is to evaluate membrane retention characteristics rather than hydrodynamics, then the impact of screen differences may be relatively minor.
Therefore, it is necessary to demonstrate that, within the scope of your study, screen type is not a critical source of variability. For example, if the main objective is to investigate the effect of membrane molecular weight cut-off (MWCO) on the retention of a target protein, and the process operates under low-shear conditions, then using an A-screen for preliminary screening may be acceptable.
However, it should be recognized that any subsequent studies involving flux, fouling, concentration time, or product aggregation may be biased due to differences in screen type.
