Asymmetric Dialysis: Truly unified single-pass ultrafiltration and buffer exchange

Asymmetric Dialysis: Truly unified single-pass ultrafiltration and buffer exchange | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 5 June 2025 V1 Latest version Share on Asymmetric Dialysis: Truly unified single-pass ultrafiltration and buffer exchange Authors : Ujwal Patil 0000-0001-6675-7899 [email protected] , Michelle Chen , Irina Ramos 0000-0002-2759-1520 , and Jon Coffman 0000-0002-4716-258X Authors Info & Affiliations https://doi.org/10.22541/au.174910937.71524550/v1 1010 views 416 downloads Contents Abstract Graphical Abstract Abstract Keywords Introduction Materials and Methods Experimental Set-up for Asymmetric Dialysis Results and Discussion Considerations for Implementation of Asymmetric Dialysis Conclusions Supplementary Material References Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Biopharmaceutical manufacturing has been using ultrafiltration (UF) and diafiltration (DF) for buffer exchange, desalting, and formulation of biologics. The legacy UF/DF is commonly a two-step batch process that is challenging to integrate into end-to-end continuous biomanufacturing. Here, we introduce asymmetric dialysis, a novel one-step continuous process that combines UF and DF. It works by utilizing asymmetric flow between the inlet and outlet of the retentate and complementary flow of the dialysate solution, achieving product concentration, buffer exchange, and salt removal using a commercially available hollow fiber device. Asymmetric dialysis can achieve product concentrations of 105 g/L (3.8x), 200 g/L (10x), 64 g/L (9.4x) starting from feed concentrations of 30 g/L, 20 g/L, and 7 g/L, respectively, with modest pressures of 6-7 psi. The interplay between feed and exchange buffer flow rates was exploited to make the process sustainable by reducing buffer consumption by 75% (25 L/kg mAb) compared to conventional batch UF/DF (100 L/kg, mAb). We successfully processed 7 kg of mAb at 20 g/L feed using 5-day asymmetric dialysis with a daily productivity of 0.8 kg/m 2 /day to product concentration of 200 g/L. These results demonstrate the potential of asymmetric dialysis a simple, sustainable, and low-cost bioprocessing technology continuous bioprocessing. Asymmetric Dialysis: Truly unified single-pass ultrafiltration and buffer exchange Ujwal Patil 1,* , Michelle Chen 1 , Irina Ramos 1 , Jon Coffman 1 1 Bioprocess Technologies and Engineering, Biopharmaceutical Development, AstraZeneca, Gaithersburg, MD, USA *Corresponding Author: [email protected] ORCID 0000-0001-6675-7899 Ujwal Patil Graphical Abstract Asymmetric dialysis, a novel one-step continuous process combining ultrafiltration and diafiltration, achieves high product concentrations (up to 200 g/L) with low pressures and reduces buffer consumption by 75% compared to conventional methods. This sustainable, low-cost technology demonstrates significant potential for integration into continuous biopharmaceutical manufacturing. Abstract Biopharmaceutical manufacturing has been using ultrafiltration (UF) and diafiltration (DF) for buffer exchange, desalting, and formulation of biologics. The legacy UF/DF is commonly a two-step batch process that is challenging to integrate into end-to-end continuous biomanufacturing. Here, we introduce asymmetric dialysis, a novel one-step continuous process that combines UF and DF. It works by utilizing asymmetric flow between the inlet and outlet of the retentate and complementary flow of the dialysate solution, achieving product concentration, buffer exchange, and salt removal using a commercially available hollow fiber device. Asymmetric dialysis can achieve product concentrations of 105 g/L (3.8x), 200 g/L (10x), 64 g/L (9.4x) starting from feed concentrations of 30 g/L, 20 g/L, and 7 g/L, respectively, with modest pressures of 6-7 psi. The interplay between feed and exchange buffer flow rates was exploited to make the process sustainable by reducing buffer consumption by 75% (25 L/kg mAb) compared to conventional batch UF/DF (100 L/kg, mAb). We successfully processed 7 kg of mAb at 20 g/L feed using 5-day asymmetric dialysis with a daily productivity of 0.8 kg/m 2 /day to product concentration of 200 g/L. These results demonstrate the potential of asymmetric dialysis a simple, sustainable, and low-cost bioprocessing technology continuous bioprocessing. Keywords Asymmetric dialysis, Diafiltration, Continuous manufacturing, Counter-current Dialysis, Single-pass bioprocessing, Ultrafiltration, Bioprocess intensification Introduction Biopharmaceutical manufacturing relies heavily on membrane-based processes to ensure the safety and quality of the drug products. In downstream purification, the membranes are most used for sterile filtration, virus retentive filtration, product concentration, buffer exchange, and pre-formulation. Depending on the application, the membrane processes are operated in either dead-end filtration (DE) or Tangential flow filtration (TFF) mode. TFF is typically operated at a relatively high flux (300 LMH) to enable high-speed movement of feed along the plane of the membrane, in turn facilitating disruption of concentration polarization and membrane fouling. Conventionally, TFF systems such as ultrafiltration (UF) and diafiltration (DF) have been used to perform concentration and buffer exchange of biologics, viz. monoclonal antibodies (mAbs). In UF/DF, the feed solution is repeatedly cycled through a membrane module in a closed loop until the desired degree of concentration and buffer exchange are attained. The inherent batch nature of the UF/DF makes it unamenable for process integration with lower flow rates encountered within continuous manufacturing. [1,2] Continuous manufacturing necessitates the use of technologies that allow the forward movement of products throughout the process without any product recirculation loops. This has prompted the introduction of single-pass tangential flow filtration (SPTFF) in membrane operations. In SPTFF, the feed is passed through the membrane module once to attain a desired concentration without product recirculation. This mode of operation relies on higher residence times obtained by using a long feed channel in combination with careful choice of feed and retentate flow rates to achieve target concentration factors compared to traditional TFF. The extended feed channel in the module is typically obtained by using a larger membrane area or adding repetitive units of identical area or deliberate (“Christmas tree”) staging of area for optimum flow channel length. [3–6] Continuous UF/DF with cassette-based SPTFF involves a multi-stage SPTFF configuration with interim dilution in co-current or a countercurrent operation, achieving a buffer exchange upwards of 99%. [7–9] Even though the counter-current approach affords a lower buffer utilization, the counter-current system is accompanied by a higher degree of operational complexity ( Figure 1 A). Unlike convective approaches that become flux-limited due to polarization effects, diffusive approaches in the form of counter-current dialysis have also been reported. The dialysis approach is operationally simple yet requires a pre-concentration step to achieve efficient buffer usage. [10] To date, SPTFF-based UF and DF operations have been distinctively separate and most often operated in a connected manner to make them amenable to continuous manufacturing. Recently, a proof-of-concept simultaneous UF and DF operation was proposed and demonstrated using a 3D-printed SPTFF prototype with limited success. [11,12] In our earlier work, we discussed the design and framework for integrated and continuous biomanufacturing, highlighting the capabilities, complexity, and sophistication required for commercial implementation. [1,2] Here, we report the use of commercially available hollow fiber membrane cartridges as a platform for performing simultaneous one-step continuous ultrafiltration, buffer exchange, and formulation of biotherapeutics. The innovative, truly unified technology discussed in this work, referred to as Asymmetric Dialysis, is capable of 10-fold product concentration and buffer exchange in a single pass. The term ”asymmetric” refers to the intentional mismatching of the entry and exit flow rates in the system and should not be confused with the characteristics of membrane pores. The adoption of a hollow-fiber format lowers spatial requirements, compared to flat-sheet cassette configurations. We evaluated the suitability and effectiveness of various hollow fiber devices for asymmetric dialysis, along with strategies for process intensification and minimizing buffer consumption. The goal was to strike a balance between spatial economy, bioprocess readiness, ease of step development, and process performance. This work also discusses considerations for the commercial implementation of asymmetric dialysis. Our results demonstrate that asymmetric dialysis can perform simultaneous concentration and buffer exchange of monoclonal antibodies (mAbs) in truly continuous operation in a simple, sustainable, and low-cost manner. not-yet-known not-yet-known not-yet-known unknown Figure 1 Comparison of legacy Ultrafiltration/Diafiltration (UF/DF) methods with hollow-fiber asymmetric dialysis. Unlike batch and continuous UF/DF (A), Asymmetric dialysis (B) performs simultaneous concentration and buffer exchange in a single pass. The feed containing mAb is pumped into the hollow-fiber cartridge using pump P1. The concentration factor along the lumen side of the cartridge is modulated using pumps P2 and P4. The fresh dialysis buffer is delivered to the shell side using pump P3. Materials and Methods All experiments were performed using either Optiflux 180NR (Fresenius, USA) hemodialyzer or Minikros 0.16 m 2 (Repligen, USA, Cat# S04-E030-05-N) bioprocess cartridge. The properties of the hollow fiber cartridges used in this study are noted in . A feed containing mAb A or B was adjusted to pH 5.0 using 50 mM sodium acetate, 200 mM sodium chloride, pH 5.0 buffer (conductivity, 21+2 mS/cm) to mimic the cation exchange mAb elution pool. The mAb concentration in the feed was adjusted between 7-30 g/L. All the experiments discussed in this work were performed with 20 mM histidine/histidine HCl, pH 5.9 + 0.1, as the dialysis buffer. Table 1 Properties of Hollow Fiber Cartridges Optiflux 180NR- hemodialyzer Fresenius Polysulfone 190 45 30 Minikros - bioprocess hollow fiber Repligen Polyethersulfone 500 200 25 Experimental Set-up for Asymmetric Dialysis A hollow fiber cartridge was placed in a vertical position and the feed was introduced from the bottom into the lumen-side port using pump P1 (as shown in Figure 1B ). The shell-side port (shell-outlet) closest to the feed port was connected to pump P2, while the distal shell-side port (shell-inlet) and lumen-side port (lumen-outlet/retentate) were connected to pumps P3 and P4, respectively. The dialysis buffer was introduced into the shell side using pump P3, while pump P2 controlled the flow rate at the shell outlet. All the hollow-fiber cartridges were operated in counter-current mode. The final concentrated and buffer-exchanged product (retentate) was collected at the lumen outlet. Before operation, the peristaltic pumps (P1, P2, P3, and P4) were calibrated by timing the collection of a known volume of deionized water using a digital balance, using the appropriate pump heads and tubing length that fit the operation. For a typical 4-pump setup using a 1.8 m 2 Optiflux 180NR hemodialyzer, pump flow rates were adjusted to allow simultaneous product concentration and buffer exchange as follows; the feed flow (P1) was 20 ml/min (0.7 L/m 2 /h or 0.7 LMH) and the pump P4 was adjusted to 5 mL/min to allow retentate flow control and achieve a 4x concentration factor along the hollow fiber membrane. Pressures were monitored using sensors before and after the inlet/outlet ports. All solutions used in the experiment were filtered through a 0.22 μm PES filter (VWR, USA, Cat# 10040-468). The shell and lumen compartments of the hollow fiber cartridge were flushed with dialysis buffer before the experiment. On the shell side, the pumps P2 and P3 were adjusted to 60 mL/min and 45 mL/min, respectively. The Optiflux 180NR hemodialyzers have a 100 mL hold-up volume therefore a steady state was defined as the passage of 2.5 times the holdup volumes through the cartridge. The product (retentate) concentration was measured after allowing the system to reach a steady state. For the experiments involving bioprocess hollow fiber, a feed flow rate was chosen to yield a feed flux that matched the hemodialyzer experiments. All experiments were performed at room temperature. The impurity removal performance of asymmetric dialysis was investigated using vitamin B 12 (Sigma-Aldrich USA, Cat# V2876) as a model impurity in the feed. All experiments were performed in single-pass mode with no recirculation. Samples were collected periodically from the lumen outlet for offline pH and conductivity (Mettler Toledo, USA). Mab and B 12 concentrations were measured using SoloVPE (Repligen, USA) at 280 nm and 550 nm, respectively. The histidine analysis was performed using Acclaim Trinity P1 column (ThermoFisher Scientific, USA, Cat# 075563). For legacy UF processes, the degree to which the volume of a solution has been reduced (concentration) is indicated by the volumetric concentration factor ( vcf ), and the extent of buffer consumption in DF is expressed as diafiltration volume (DV). To succinctly denote the simultaneous concentration and buffer exchange afforded by asymmetric dialysis, we introduce the term asymmetry factor (AF) expressed as \(\text{AF}_{\alpha^{{}^{\prime}}}^{\text{vcf}}\). Herein, vcf is the volumetric concentration factor along the lumen, calculated as the ratio between feed flow and retentate flow; α’ is the buffer consumption calculated as the ratio between dialysis buffer flow and retentate flow. For example, a process with a concentration factor of 10 and α’ of 22.5 is depicted as \(\text{AF}_{22.5}^{10}\); for additional examples and mass balances for inputs and outputs, please refer to SI Table 1 in the supplementary information. Results and Discussion To determine the appropriate hollow fiber for continuous bioprocessing, the asymmetric dialysis performance was evaluated using two types of hollow fiber cartridges, (1) Optiflux F180NR (1.8 m 2 ) hemodialyzer and (2) Minikros (0.16 m 2 ) bioprocess hollow fiber, with mAb A feed mimicking a cation-exchange chromatography (CEX) elution pool (50 mM sodium acetate buffer, 200 mM sodium chloride, pH 5.0). Figure 2 shows the single-pass concentration and buffer exchange of feed containing 7, 20, and 30 g/L mAb at a feed flux of 0.7 LMH for 3-pump setup without deliberate retentate flow control (P4). Shell-side flow rates were adjusted to target \(\text{AF}_{22.5}^{10}\)(for 7 and 20 g/L) and \(\text{AF}_{11.2}^{4}\) (for 30 g/L). The product concentrations obtained with the hemodialyzer, and bioprocess hollow fiber cartridges were comparable and yielded average product concentrations of 60 g/L, 160 g/L, and 110 g/L (Figure 2A). The hemodialyzer resulted in concentration factors of 8.5x, 8.2x, and 3.7x, whereas concentration factors of 8.5x, 7.4x, and 3.8x were achieved with bioprocess hollow-fiber cartridges. As shown in Figure 2B/C, both membranes were able to reduce feed conductivity from 20 mS/cm to product conductivity of 1.5 mS/cm. Similarly, for buffer exchange, the overall product pH after asymmetric dialysis was found to be 6.0 + 0.2 for both membranes, as targeted. The transmembrane pressure (TMP) across both cartridges was less than 5.0 psi, with the highest TMP of 4.0 psi for the hemodialyzer for the product concentration of 160 g/L. Interestingly, for 7 and 20 g/L feed, both cartridges fell short of their intended concentration target of 10x by 15%. We believe that at higher (> 6x) concentration factors the lack of flow control on the retentate limits the maximum attainable concentration factor across the cartridge. To overcome this limitation, all subsequent investigations included the use of a retentate pump that would establish enough TMP to overcome the membrane resistance enabling consistent and tunable product concentration. Even though both membrane units represent a general hollow fiber format, key differences in fiber dimensions, shape, packing, and shell-side compartments could make them perform differently when operated in counter-current flow (Table 1). For instance, the thinner fibers (~45 μm) in hemodialyzers facilitate better mass transport and are typically rated for low-pressure (< 10 psi) operation. In contrast, fibers for the bioprocess cartridges have higher wall thickness (~200 μm) required to sustain higher pressures (10-20 psi) in UF. In addition, the undulations on fibers in hemodialyzers from micro-crimping contribute to enhanced mass transfer compared to straight cylindrical fibers. [13,14] Unlike the bioprocess cartridges, shell-side compartments of the hemodialyzers are equipped with an annular flow distributor or flow baffles which mitigate the flow maldistributions on the shell-side that can severely impact the buffer exchange performance [15–19] . Historically, bioprocess hollow fibers were designed for UF, not dialysis, and hence have not benefitted from the technological improvements made in hemodialyzers over the four decades. [20] To take advantage of cumulative advancements, we opted to utilize hemodialyzers in our subsequent studies. Figure 2 Asymmetric dialysis performance of hemodialyzer (black) and bioprocess hollow fiber (grey) cartridges. (A) Product concentrations (y-axis) were obtained for respective mAb A feed concentrations (x-axis). The dotted lines indicate intended target concentration factors of 10x, 10x, and 4x, for 7, 20, and 30 g/L respectively. (B) Product conductivity and (C) pH after asymmetric dialysis for respective mAb A feed concentrations. To explore opportunities to intensify asymmetric dialysis for higher product throughput, we assessed the effect of varying feed concentration and feed flux on overall process performance. All the subsequent studies were performed in the retentate flow control mode using a peristaltic pump (P4) to attain the desired concentration factor. The feed flux and concentration represent practically relevant productivity scenarios in integrated and continuous manufacturing. [1,2,21] In this study, mAb A feed at 7 and 20 g/L in 50 mM sodium acetate, 200 mM NaCl, pH 5, was processed at \(\text{AF}_{22.5}^{10}\) for feed flux ranging from 0.7 to 3.5 LMH (Figure 3). For example, for feed flux of 0.7 LMH, the feed and retentate flow rates were 20 mL/min and 2 mL/min, respectively; the shell-inlet and outlet flow rates were 45 mL/min and 63 mL/min, respectively. The inlet and outlet flow rates for the lumen and shell side were adjusted to ensure flow totalization of the system. As shown in Figure 3A, 7 g/L mAb feed with \(\text{AF}_{22.5}^{10}\) did not impact buffer exchange performance as the pH and conductivity of the product were comparable to the dialysis buffer, 20 mM histidine pH 5.9 and 1.5 mS/cm. For 20 g/L mAb feed, product conductivity was comparable to the dialysis buffer, whereas the product pH was overshot by 0.2 pH units compared to the dialysis buffer. The dependency of pH offset on target product concentration when compared between 7 and 20 g/l for the same \(\text{AF}_{22.5}^{10}\) can be attributed to the Donnan effect typically observed at mAb concentrations above 120 g/L. [22] We investigated the excipient (histidine) concentration in the product from the dialysis buffer. From Figure 3B, it can be observed, that at 120 g/L the product contains (25% lower) 15 mM histidine compared to the dialysis buffer (20 mM histidine). As the product concentration increases from 160 g/L to 200 g/L the extent of the drift is less prominent, stabilizing at 13 mM histidine. The drift in targeted pH and excipient concentration relative to the dialysis buffer in asymmetric dialysis are comparable to the ones that have been reported earlier in UF/DF. [22–24] This drift is a product of the Donnan and volume exclusion effects due to unequal partitioning of charged solutes across the membrane, leading to unequal distribution of electrolytes. [25,26] The drift in the product histidine can be compensated by increasing the histidine concentration in the dialysis buffer, like the strategy commonly employed in UF/DF. [27] For both 7 g/L and 20 g/L feeds, the asymmetric dialysis was able to achieve the intended concentration factor of 10 for fluxes 0.7 LMH and 1.7 LMH. However, at 3.5 LMH, the system suffered over-pressurization (TMP >10 psi) and was ultimately unable to achieve the desired concentration factor for both feed concentrations. This is likely due to critical-flux limitations at higher feed flux exacerbated by the gel layer formation ultimately leading to increased membrane resistance. For the remainder of feed fluxes, the steady-state TMP values for 7 g/L feed ranged from 2 to 3 psi and 5 to 7 psi for 20 g/L and remained consistent (+10%) throughout the operation (Figure 3B). The benign TMPs observed in asymmetric dialysis eliminate the need for engineered provisions such as stainless-steel manifolds and heavy-duty pumps typically employed in inherently high-pressure (25 psi) UF/DF and SPTFF operations. Interestingly, despite being a closed system with pumps on each outlet, asymmetric dialysis can operate without catastrophic pressurization at a peak flux of 1.7 LMH or lower. This is likely due to inherent flow slippage in peristatic pumps that allows the system to self-balance. The flux excursion investigation demonstrates the ability of asymmetric dialysis to process 1.5 kg mAb per (1.8 m 2 ) membrane unit per day at 1.7 LMH feed flux for 20 g/L mAb feed and 200 g/L product. In addition, the tunability between 0.7 LMH to 1.7 LMH is beneficial for accommodating changes in material throughput in an end-to-end continuous process and can be particularly useful in the event of process upsets. Figure 3 Performance of intensified asymmetric dialysis using hemodialysis cartridge (Optiflux 180NR) (A) Product pH (primary y-axis, circle) and conductivity (secondary y-axis, square) after asymmetric dialysis of 7 g/L (black) and 20 g/L (white) mAb feed. (B) Histidine concentration in the AD product using 20 g/L feed at various target product concentrations compared with His concentration in the dialysis buffer (dashed line, 20 mM). (C) TMP profile for 20 g/L feed at 1.7 LMH feed flux and 10x concentration factor. For high-concentration formulations, it is common for a UF/DF process to be accompanied by a secondary UF to achieve the target concentration (>200 g/L) that can significantly increase both cost and process complexity. In addition, UF/DF necessitates feed flow tuning to counteract the flux challenges at elevated product viscosity and pressures that could exacerbate risks of protein aggregation due to shear stress and recurrent pump circulations. Owing to its low feed flux and single-pass mode, asymmetric dialysis is a low-shear process and can tolerate retentate viscosities up to 120 cP for high-concentration formulations without a deliberate process and buffer optimization (SI Figure 2). The ability to process high-concentration viscous products presents an additional benefit in reducing Scope 3 emissions from bulk-liquid freight to fill-finish sites. The changes in membrane performance were assessed using Normalized Water Permeability (NWP) to measure the degree of fouling when compared with an unused membrane. The membranes typically retained 70% NWP after use (SI Figure 1). \begin{equation} \alpha׳=\frac{q_{D}}{q_{P}}\nonumber \\ \end{equation} \(q_{P}=\frac{q_{F}}{\text{CF}}\); The Process Mass Intensity (PMI) is the volume of buffer consumed per kilogram of the product processed can be given as \begin{equation} PMI\ (L/kg)=\frac{q_{D}}{q_{P}\times\ C_{p}}\nonumber \\ \end{equation} where q F , feed flow; CF , concentration factor; C p , product concentration at retentate (kg/L). Figure 4 Effect of decreasing α’ values on buffer usage and exchange performance at 1.5 LMH feed flux. (A) Buffer consumption per kg of mAb (y-axis) for a 20 g/L mAb A feed processed using asymmetric dialysis to target product concentration of 200 g/L at various α’. (B) Product pH (solid gray circles, primary y-axis) and conductivity (solid black squares, secondary y-axis) after asymmetric dialysis of 20 g/L feed at various α’. (C) Histidine concentration in the product after asymmetric dialysis at various α’. (D) Vitamin B 12 concentration in the 225 g/L asymmetric dialysis product stream using 1.8 m 2 hemodialyzer operated at 1.5 LMH feed containing 21 g/L mAb with 4.9 g/L vitamin B 12 processed with\(\text{AF}_{5}^{10}\). Figure 4A shows the buffer consumption corresponding to AF values ranging from \(\text{AF}_{22.5}^{10}\) to \(\text{AF}_{3}^{10}\) where the vcf is held constant at 10 and α’ is varied between 3 and 22.5, with 20 g/L mAb A feed at 1.5 LMH feed flux, dialyzed with 20 mM histidine buffer, pH 6.0. For an α’ of 22.5, the resulting dialysis buffer consumption was 110 L/kg of mAb, which is as buffer-intensive as legacy UF/DF. In comparison, optimized α’ values of 5 and 3 demonstrated buffer consumption of 25 L/kg mAb and 15 L/kg mAb, respectively. As shown in Figure 4B, reducing α’ to as low as 5 did not affect the buffer exchange performance with no noticeable difference in product pH and conductivity for α’ ranging between 22.5 to 5. Further attempts to lower buffer utilization, α’=3, resulted in poor buffer exchange performance. This indicates the possibility of using a lower α’ (< 3) for applications such as intermediate pH or conductivity conditioning between unit operations that do not necessitate 99% exchange. Interestingly, lowering the α’ beyond 22.5 had a negligible impact (10%) on product histidine concentration (Figure 4C). This indicates that the buffer exchange performance is governed by the Donnan equilibrium and is largely unaffected by the availability of the fresh dialysis buffer. The effect of the reduced α’ on impurity removal was investigated by spiking the (20 g/L) mAb feed with 4.9 g/L vitamin B 12 as a model impurity. Figure 4D shows the degree of vitamin B 12 removal at α’ of 5 and 1.5 LMH feed flux. As the system reached a steady state (~70 min) the vitamin B 12 in the product was reduced to 0.015 g/L indicating a 99.7% removal with no impact on buffer exchange and salt removal (SI Table 2). The asymmetric dialysis, at α’=5, uses 75% less buffer compared to batch UF/DF (performed at 60 g/L) and 70% less buffer than continuous cc-SPTFF. [8] In addition, cc-SPTFF has an elaborate engineering design, necessitating break-vessels, and complex automation, while asymmetric dialysis is more straightforward to establish and simpler to automate. Interestingly, the estimated buffer consumption in asymmetric dialysis reduces proportionally to the increased concentration factor (SI Figure 3). However, higher concentration factors for 20 g/L mab feed or higher would likely require corresponding adjustment in flux to limit system over pressurization (> 10 psi) while still maintaining the buffer savings. Overall, the results demonstrate that asymmetric dialysis is a more efficient alternative to legacy and recent technologies for concentration and buffer exchange. Based on the flux and buffer consumption (α’) parameters identified above, we evaluated the endurance of the asymmetric dialysis technology for a continuous and uninterrupted operation at a feed flux of 1.7 LMH and α’ of 5. As shown in Figure 5A, feed containing 20 g/L mAb A, 50 mM sodium acetate buffer, and 200 mM sodium chloride at pH 5 was processed through 1.8 m 2 hemodialyzer with a\(\text{AF}_{5}^{10}\) to yield a product concentration of 200 g/L. The product concentration measured at the retentate remained largely within 5% of the desired target, 200 g/L (Figure 5B). Similarly, the pH and conductivity of the product were maintained at 6.1 + 0.1 and 1.6 + 0.1 mS/cm throughout the 5-day run (Figure 5C). The process operated continuously without requiring intermittent chemical or physical cleaning and flux alterations throughout the operation. Over 120 hours, asymmetric dialysis processed 7 kg mAb with membrane loading of the NWP when compared to the unused membrane indicating the potential to utilize the membrane beyond 5 days. Also, indicated by the constant TMP of 5 psi observed from steady-state through the 5 days of operation. The process demonstrated an overall product recovery of 99%. Unlike UF/DF, the continuous nature of asymmetric dialysis eliminates the need for product chase; in turn, product losses for the runtime of 24h or higher are about 1%. To note, the process was operated without any automation or human intervention after attaining a steady state, this highlights the ease of use in a lab-scale environment. Figure 5 Continuous 120-hour (5-day) asymmetric dialysis operated at 1.7 LMH feed flux. (A) Schematic of the setup with the flow rates used for lumen and shell-side pumps to process 20 g/L mAb A feed containing 50 mM sodium acetate buffer, and 200 mM sodium chloride at pH 5 using a 1.8 m 2 hemodialyzer cartridge. (B) Retentate mAb concentration (secondary y-axis) and moving average of TMP (primary y-axis) and profile over 5 days. Post-steady state (0-2 h), the pressure profile remains largely unchanged with an average TMP of 5 psi. The retentate mAb concentrations obtained ranged between 190-208 g/L, well within an intended target concentration range of 190-210 g/L (grey rectangle). The floating x-axis indicates the amount of mAb processed over 5 days with daily productivity of 1.5 kg/day. (C) pH (square) and conductivity (triangle) profile of the product throughout the 5-day run Considerations for Implementation of Asymmetric Dialysis The path to implementation of any new technology is governed by technical feasibility, robustness, integration complexity, regulatory compliance, and end-user acceptance. In the technology life cycle, from idea to adoption, these factors evolve through stages of development, testing, and validation. Technology readiness levels provide a structured framework to assess and communicate the maturity and readiness of a technology for commercial deployment. [28] For asymmetric dialysis, we evaluated the bioprocess readiness level tool created by NIIMBL and scored up to BRL5. While it is beyond the scope of this manuscript to holistically address the requirements for implementing this technology, we aim to highlight key actions that would accelerate its successful adoption. Table 2 Process economic comparisons for the UF/DF, cc-SPTFF, and Asymmetric dialysis. UF/DF 60 g/L 110 CIP 8000 1.4 5700 cc-SPTFF † 200 g/L 60 CIP 40000 6 6700 Asymmetric dialysis 200 g/L 25 gamma/e-beam irradiated 50 4 (5-day) 15 † For 4-stage cc-SPTFF process at 10 LMH feed flux and 20 g/L feed The process economic considerations of legacy UF/DF, SPTFF, and asymmetric dialysis, are juxtaposed in Table 2 for the respective target product concentrations. Amongst the three technologies, hemodialyzers used in asymmetric dialysis offer extremely low Cost of Goods manufacturing (CoGm) at $15/kg, mAb. These cost savings are even more impactful towards clinical manufacturing campaigns lowering the overall cost of drug development. Whereas, to improve the CoGm of the TFF membranes, they would need to be reused 100 times. However, each reuse incurs additional costs for labor, CIP, validation, and energy. Although theoretically feasible, this approach is not economically practical. The hemodialyzers offer a “plug and play” design for single-use and are provided in a pre-sterilized format, which would greatly facilitate the rapid replacement of cartridges as required. In addition, the design and simplicity of asymmetric dialysis allow a compact setup yielding a smaller GMP footprint combined with space savings due to the reduction in buffer storage area compared to legacy UF/DF (SI Figure 4). Hemodialyzers, while economically attractive, require accommodations to be assimilated for bioprocess use. These cartridges do not employ conventional bioprocess connectors. This challenge can be mitigated by custom flow kits, that can be assembled aseptically with membrane units, as a bridging solution until bioprocess-ready units become available. Millions of hemodialyzers are produced globally every year, of which the biopharma sector will be a relatively minor consumer. Although this ensures a consistent supply, hemodialysis devices are classified as FDA class II products, and their transition to bioprocessing requires appropriate certifications. For example, bioprocess vendors usually offer extractable and leachable evaluations, and the lack of bioprocessing-ready units necessitates the creation of supporting documentation based on the available USP and BioPhorum Operations Group (BPOG) guidelines. [29] Figure 6 Schematic for the flow ratio-based feedback control of asymmetric dialysis indicated by dashed arrows The control strategy for asymmetric dialysis is anchored in flow ratio control. The flow ratio control of the pumps is informed by continuously monitoring process parameters: flow rate, protein concentration and TMP. The retentate pump (P4) can serve as the concentration controller to attain a consistent product concentration (Figure 6). This can be achieved by feedback control of the P4 and q P , measured in real-time using an inline mass flow meter downstream to P4. During startup, until the system is in a steady state, the flow diversion valve will divert the retentate flow to waste, once the UV setpoint is reached, the retentate flow is diverted to the product collection vessel. We believe this type of control is much simpler than the multistage TFF approaches discussed elsewhere. Depending on the material accumulation rate from the preceding operation, the asymmetric dialysis feed flow can be adjusted between 0.1 to 1.7 LMH, which then dynamically adjusts the P4 and P3 to maintain the desired\(\text{AF}_{\alpha^{{}^{\prime}}}^{\text{vcf}}\). The spent dialysate flow P2 is set to totalize the net outflow from the system. The feed flux modulation allows the system to respond to process upsets or delays in the preceding operation and maintain a desired volume level in the feed tank. The currently available hemodialyzers would provide a more than adequate capacity (0.8 kg/m 2 /day) for typical clinical-scale end-to-end continuous manufacturing involving a 1000L perfusion bioreactor. However, for commercial-scale, larger membrane units need to be developed as the existing hemodialyzer membrane areas are scaled to human use with 2.5 m 2 being the maximum. This limitation can be addressed using a scale-out approach by operating multiple hemodialyzers either in series or parallel mode. The scale-out approach could also open opportunities for asymmetric dialysis to be employed in fed-batch operations where a higher membrane area is required over a shorter processing time. Conclusions In this work, we proposed a novel technology that can replace UF/DF operation in bioprocessing. We demonstrated practically relevant capabilities of asymmetric dialysis for continuous unified concentration and buffer exchange using commercially available hemodialyzers. The intensified process can operate at the highest feed flux of 1.7 LMH with mAb productivity of 0.8 kg/m 2 /day with minimal membrane fouling. Unlike UF/DF, where 10–100 pump passes are required, the single-pass operation in our approach could result in improvements in product quality, especially for shear-sensitive products. We successfully processed 7 kg of mAb with consistent salt removal and buffer exchange performance while maintaining a product concentration in a narrow range (within 5% of target) for 5 days of continuous operation. Additional engineering hardware, PATs, and automation will be required to make the technology GMP-ready. In addition, we lowered the buffer consumption to 25 L/day which is a 75% reduction compared to UF/DF. Hemodialyzers offer a single-use solution that is extremely cost-effective ($15/kg, mAb) and with a membrane capacity upwards of 4 kg/m 2 . We have highlighted microstructural (fiber thickness) and macrostructural (cartridge design) properties of the hemodialyzers that are lacking from the bioprocess hollow fibers in the market. The hemodialyzers can be qualified for bioprocess use as a bridging solution until the bioprocess equivalent alternatives with the desired properties become available. To summarize, the simplicity and sustainability benefits presented by asymmetric dialysis can be materialized through a concerted change management effort by addressing the technical, engineering, and regulatory nuances. We envision this technology to be forward and backward compatible as it can be applied to continuous as well as legacy batch operations by scaling out for appropriate membrane areas. Beyond mAbs, our generalized approach can be applied to modalities such as fusion molecules, enzymes, nanoparticles, etc. Besides the advantages, opportunities for further improvements such as the need for small-scale down models to facilitate process development efforts were also identified. To our knowledge, this is the first technology that achieves single-pass simultaneous product concentration and buffer exchange in a continuous format. We believe that the benefits and simplicity of asymmetric dialysis make it a successor to state-of-the-art UF/DF technologies and a key element in end-to-end continuous biomanufacturing. Supplementary Material File (image10.emf) Download 97.62 KB File (image3.emf) Download 77.28 KB File (image4.emf) Download 88.04 KB File (image7.emf) Download 114.81 KB References 1. [1] J. Coffman, K. Bibbo, M. Brower, R. Forbes, N. Guros, B. Horowski, R. Lu, R. Mahajan, U. Patil, S. Rose, J. Shultz, Biotechnol Bioeng 2021 , 118 , 3323. [2] J. Coffman, M. Brower, L. Connell‐Crowley, S. Deldari, S. S. Farid, B. Horowski, U. Patil, D. Pollard, M. Qadan, S. Rose, E. Schaefer, J. Shultz, Biotechnol Bioeng 2021 , 118 , 1735. [3] E. Madsen, J. Kaiser, U. Krühne, M. Pinelo, Sep Purif Technol 2022 , 291 , 120949. [4] C. J. Yehl, A. L. Zydney, J Memb Sci 2020 , 595 , 117517. [5] E. M. Goodrich, D. M. Bohonak, P. W. Genest, E. Peterson, Approaches to the Purification, Analysis and Characterization of Antibody-Based Therapeutics 2020 , 137. [6] M. G. Jabra, C. J. Yehl, A. L. Zydney, Biotechnol Prog 2019 , 35 , 6. [7] J. Rucker-Pezzini, L. Arnold, K. Hill-Byrne, T. Sharp, M. Avazhanskiy, C. Forespring, Biotechnol Bioeng 2018 . [8] A. M. K. Nambiar, Y. Li, A. L. Zydney, Biotechnol Bioeng 2018 , 115 , 139. [9] R. T. Kurnik, A. W. Yu, G. S. Blank, A. R. Burton, D. Smith, A. M. Athalye, R. van Reis, Biotechnol Bioeng 1995 , 45 , 149. [10] C. J. Yehl, M. G. Jabra, A. L. Zydney, Biotechnol Prog 2019 , 35 , 1. [11] R. Tan, M. Franzreb, Biotechnol Bioeng 2020 , 117 , 654. [12] R. Tan, M. Franzreb, Sep Purif Technol 2022 , 282 , 119987. [13] C. Ronco, A. Brendolan, C. Crepaldi, M. Rodighiero, P. Everard, M. Ballestri, 2000 , 23 , 601. [14] J. K. Leypoldt, A. K. Cheung, T. Chirananthavat, J. F. Gilson, C. D. Kamerath, R. B. Deeter, ASAIO Journal 2003 , 49 , 81. [15] I. Noda, D. G. Brown-West, C. C. Gryte, J Memb Sci 1979 , 5 , 209. [16] M. J. Costello, A. G. Fane, P. A. Hogan, R. W. Schofield, J Memb Sci 1993 , 80 , 1. [17] R. A. Ward, J. W. Idoux, H. Hamdan, R. Ouseph, T. A. Depner, T. A. Golper, Clinical Journal of the American Society of Nephrology 2011 , 6 , 2235. [18] J. Lemanski, G. G. Lipscomb, J Memb Sci 2002 , 195 , 215. [19] P. A. Hardy, W. R. Clark, M. Park, D. Gao, 2003 , 125 , 481. [20] N. Said, W. J. Lau, Y. Ho, S. K. Lim, M. Nidzhom, Z. Abidin, 2021 , 1. [21] I. Ramos, N. Sharda, R. Villafana, K. Hill‐Byrne, K. Cai, J. Pezzini, J. Coffman, Biotechnol Bioeng 2023 , 120 , 1869. [22] J. M. West, H. Feroz, X. Xu, N. Puri, M. Holstein, S. Ghose, J. Ding, Z. J. Li, Biotechnol Bioeng 2021 , 118 , 2293. [23] V. Hebbi, S. Roy, A. S. Rathore, A. Shukla, Sep Purif Technol 2020 , 238 , 116392. [24] R. Ambrožič, D. Arzenšek, A. Podgornik, Biotechnol Bioeng 2021 , 118 , 633. [25] J. Abel, A. Kosky, N. Ball, H. Bacon, R. Kaushik, G. R. Kleemann, J Pharm Sci 2018 , 107 , 1296. [26] G. R. Bolton, A. W. Boesch, J. Basha, D. P. LaCasse, B. D. Kelley, H. Acharya, Biotechnol Prog 2011 , 27 , 140. [27] F. Miao, A. Velayudhan, E. Dibella, J. Shervin, M. Felo, M. Teeters, P. Alred, Biotechnol Prog 2009 , 25 , 964. [28] S. B. Kedia, J. C. Baker, R. G. Carbonell, K. H. Lee, C. J. Roberts, J. Erickson, J. E. Schiel, K. Rogers, G. Schaefer, S. Pluschkell, Biotechnol Bioeng 2022 , 119 , 3526. [29] W. Ding, G. Madsen, E. Mahajan, S. O’Connor, K. Wong, Pharm Eng . 2014, pp. 1–11. Google Scholar Information & Authors Information Version history V1 Version 1 05 June 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords asymmetric dialysis continuous manufacturing counter-current dialysis diafiltration single-pass bioprocessing ultrafiltration Authors Affiliations Ujwal Patil 0000-0001-6675-7899 [email protected] AstraZeneca R&D Gaithersburg View all articles by this author Michelle Chen AstraZeneca R&D Gaithersburg View all articles by this author Irina Ramos 0000-0002-2759-1520 AstraZeneca R&D Gaithersburg View all articles by this author Jon Coffman 0000-0002-4716-258X AstraZeneca R&D Gaithersburg View all articles by this author Metrics & Citations Metrics Article Usage 1010 views 416 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Ujwal Patil, Michelle Chen, Irina Ramos, et al. Asymmetric Dialysis: Truly unified single-pass ultrafiltration and buffer exchange. Authorea . 05 June 2025. DOI: https://doi.org/10.22541/au.174910937.71524550/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174910937.71524550/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a0038347793f0db4',t:'MTc3OTUzMzU2NA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();