Design of Highly Retentive and Robust Virus Filters through Fundamental Understanding Nigel Jackson1, Engin Ayturk2, Stanley Kidd2 1. Pall Europe Ltd, Portsmouth UK; 2. Pall Corporation, Port Washington, USA INTRODUCTION COST EFFICIENT VIRUS FILTER DESIGN Understanding of the fundamental mechanisms governing performance of a membrane is key to realising the design of a high performance next generation product. This study demonstrates the original internal polarization model derived for first generation virus removal filters and its effective utilization to develop highly retentive and robust next generation virus filters. MATERIALS AND METHODS Retention experiments were carried out using PP7 bacteriophage (model parvovirus) derived from ATTC samples and stored in tryptone soya broth (TSB). Titres were determined using a plaque assay method. PP7 was fluorescently labelled using an Alexa Fluor® 488 protein labeling kit (Invitrogen, OR, USA) after dialysis into PBS. Confocal microscopy images were generated using a Leica SP5 confocal microscope (Leica Microsystems GmbH, Germany). XY Images were stacked in the Z-plane to generate a 3-D image of the membrane using ImageJ software (NIH, US). Full details of all methods are published[1]. All other test solutions and filtration conditions are as described. The lack of model fit for single-layer can be explained by hindered convection of the virus through narrower pores in the Pegasus SV4 membrane. A revised model is shown in Figure 2 taking into account a hindered convection value equivalent to 20L.m-1 delay, based on the relatively slow hindered velocity (Equation 4)[3]. uH = hindered velocity, uP = pore velocity Equation 4: KS = convective resistance coefficient KS uH = uP 2K T (2 ) K S , KT , = f( ) KT = diffusive resistance coefficient Φ = partition coefficient, λ = dvirus / dpore In order to generate a significantly high hindrance, a virus to pore ratio of 96% is required. Figure 3 Highly Retentive Pegasus SV4 8 Further work is ongoing to evaluate the impact of the tortuous path a virus must take through a filter and the likelihood hindrance from the bulk convective flow. Figure 1 demonstrates the location of phage within the depth of the Ultipor VF DV20 membrane. Experiments carried out at different phage loading levels demonstrated that the distribution of the phage throughout the membrane did not change, however the intensity of the signal did. This led to the formulation of the internal polarization (IP) theory[1] which describes an accumulation of phage in the reservoir zone causing an increased challenge to the rejection zone and an apparent reduction in retention (Equations 1-3). Nevertheless, the new filter design approach successfully predicted highly retentive performance of the dual layer Pegasus SV4 (Figure 3) for both PP7 bacteriophage and PPV mammalian virus at high throughput. 6 LRV INTERNAL POLARIZATION 4 0 Figure 1 Confocal microscope image of Ultipor VF DV20 membrane representing a small cross section of membrane but the full membrane depth. Upstream (layer main), downstream (layer inset). PPV Mammalian Virus PP7 Bacteriophage 2 0 100 200 300 Throughput (L.m-2) PRESSURE INTERRUPTIONS FLOW RESE RVOIR RESERVOIR ZONE Robust performance of Pegasus SV4 after a 10 minute pressure interruption preceding a 10% buffer flush for both IVIG and BSA solutions is shown in Figure 4. REJECTION ZONE Overall, Pegasus SV4 demonstrates very robust performance in the presence of pressure interruptions, which is expected due to the deviation from the standard IP model and the lack of virus mobility. Parameters in the diffusion driven phenomena of pressure interruptions[4], are summarized in Table 1. It should be noted that such phenomena are a theoretical risk for all virus filters, no matter how robust, and best practice requires process control of pressure interruptions to maximise virus safety. Equation 1: ( C1 x 1 e CF S C1 = first layer filtrate concentration Equation 2: SV VR ) ( C2 x = 1 e CF S 2 Process controls are crucial for the smooth transition from protein to buffer in order to minimise the impact on retention (particularly for recombinant proteins or monoclonal antibodies). An example of such process control is as shown in Figure 5 (using an Allegro™ MVP System). SV VR ) x V ( e VR 2 SV VR C2 = second layer filtrate concentration ) Figure 4 Effect on pool LRV of a 10 minute pause and 10% buffer flush for 4% IVIG and 0.1% BSA IVIG IV VIG BSA BS SA 0 CF = feed concentration x = fraction of viruses in free solution Equation 3: () V n log VR V = volume filtered ΔLRV ΔLRV LRV n log xS + log( n!) S = sieving coefficient of the rejection zone VR = volume of reservoir zone n = number of layers The IP model (Eqns. 1-3) describes the theoretical change in filtrate virus concentrations and LRV, which is not seen for larger viruses, nor is it universal for all feed solutions. The polarization effect is strongest in the initial aliquots when the amount of virus retained in the membrane is low compared to the input concentration. This is followed by a stable LRV asymptote at higher throughputs. Theoretically a similar membrane with even higher intrinsic retention (lower xS) would generate a higher LRV at the stable asymptotic region and a very highly retentive filter. NEW VIRUS FILTER DESIGN Pegasus SV4 maximises the benefits of first generation virus filters by maintaining the strong resistance to flux decay but increasing the flow and the retention performance (Figure 2). Figure 2 Relative PP7 filtrate concentration vs Throughput for single / dual layer Pegasus SV4 1.E-04 Improved single layer intrinsic retention leads to a stable and high LRV performance combined with strong flux and resistance to flux decay. 1.E-05 Cfiltrate / C feed Achieved good agreement between the IP model fit and stable high retention of the dual layer membrane. Developed a revised model to better explain single-layer performance (Equation 4). Figure 5 Process Control of Pressure Interruption Using an Automated Virus Filter System -0.5 -1 Table 1 Key Parameters that can Affect this Phenomenon Include: Fixed Parameters Virus Size Membrane Pore Size/Distribution Low-Impact Parameters Feed Viscosity Process Temperature Virus Input Concentration Throughput High-Impact Parameters Time (duration of flow pause or pressure interruption) Magnitude of Pressure Reduction (reduction in convectice flow) • Fixed by membrane choice and virus type • Minimal variation possible within a given process • Affects both input viral load and number of difused viruses – low impact on process titer reduction • Potential to vary by orders of magnitude for a given process • Can be reduced to minimize pressure interruption using proper process controls (see Pall’s position risk mitigation strategy) see section 2 CONCLUSIONS 1.E-06 By understanding the fundamental mechanism behind previous generations of virus filters it was possible to drive improvements in virus filter development and generate a highly retentive and robust virus filter. For best practice and to maximise virus safety a robust virus filter should still be combined with a robust and controlled process. 1.E-07 0 50 100 Throughput (L.m-2) 150 Dual Layer Dual Layer IP Model Single Layer Single Layer IP Model Single Layer IP Model with Hindered Convection *BSA challenge solution prepared as per PDA guidelines[2]. REFERENCES [1] Jackson N, Bakhshayeshi M, Zydney AL, Mehta A, van Reis R, Kuriyel R (2014) Biotechnol Prog 30(4):856-63 [2] PDA Virus Filter Task Force (2005) Technical report 41. PDA J Pharm Sci Technol. 59(S-2). [3] Bungay, Brenner (1973) Int J Multiphase Flow, 1:25-26. [4] Pall publication USD2997: Pressure Interruptions (Stop/Start) During Virus Filtration: Assuring Virus Safety Using Robust Process Technology and An Appropriate Risk Mitigation Strategy. Contact: +800.717.7255 (USA) • +41 (0)26 350 53 00 (Europe) • +65 6389 6500 (Asia/Pacific) • E-mail: [email protected] • Web: www.pall.com/biopharm © 2015, Pall Corporation. Pall , Allegro, Pegasus and Ultipor are trademarks of Pall Corporation. ® indicates a trademark registered in the USA. ◆Biosolve is a trademark of BioPharm Services, Chesham, UK. 5/15, GN15.6285
© Copyright 2026 Paperzz