OR WAIT null SECS
Lifetime studies for membrane reuse remains a critical issue for biotech companies producing therapeutic and diagnostic products. Two approaches commonly used for establishing membrane lifespan are prospective validation and concurrent validation. Creation and qualification of the scale-down model, and performance of small-scale reuse studies to support prospective validation, is time-consuming and costly. However, if a high likelihood of cleaning failure exists, performing these studies may be essential and may prevent loss of manufacturing lots at scale. For certain cases, concurrent validation may be more appropriate and performing an identical testing regimen in the form of routine in-process testing may provide the same, if not higher, levels of assurance for lifetime membrane performance. A production schedule may be significantly influenced when additional sampling and testing are required between lots. Therefore, a cost–benefit analysis should be performed to determine which validation approach, prospective or concurrent, is more appropriate and also how many reuses should be targeted.
Membrane filtration steps such as ultrafiltration or diafiltration (UF/DF) and microfiltration are extensively used in biotech processes. The objective of these steps may be clarification, concentration, or diafiltration. The popularity of filtration steps stems from their robustness, ease of process development, and scale up. Membrane filters, however, can be expensive. For example, UF/DF membranes required to process a single batch corresponding to a 10-kiloliter cell-culture process cost between $100,000 and $1 million.1 Therefore, from a process economics point of view, it is important to reuse membrane filters for multiple cycles before replacing them. The benefits of reuse include cost savings for the filters themselves, reduced changeover times and the resulting operational flexibility, and reduced labor and materials costs.
Cleaning procedures, and possibly steam sanitization procedures, are routinely performed before membrane use. Depending on the specific application and process involved, membrane performance and functionality may be adversely affected over time.
Potential negative effects of long-term reuse include physical deterioration of the membranes or associated hardware due to repeated exposure to cleaning agents at elevated temperatures and pressures. These effects may influence the filtration performance of the membranes, which, in turn, may affect product quality. Moreover, microbial contamination may occur in membrane systems that cannot be cleaned by routine operations. Such buildup may result in carryover of impurities or product from one lot to subsequent lots. Thus, lifetime studies for membrane reuse are necessary to ensure that membrane functionality does not deteriorate over time to the point where it affects the process and the product.
Anurag S. Rathore, PhD
The topic of reuse of membrane media, unlike the topic of reuse of chromatographic media, has not received much attention in the literature. The present article is the ninth in BioPharm International's "Elements of Biopharmaceutical Production" series, and it presents approaches toward establishing and demonstrating lifetimes for membrane media.
An appropriate subset of parameters should be chosen when planning a reuse study. These parameters should be based on the purpose of the intended application. This section briefly discusses some important process parameters commonly used to assess the performance of a membrane step.
Normalized water permeability (NWP) is perhaps the most commonly used performance parameter for monitoring the cleanliness of a UF/DF membrane. NWP uses water to measure the permeability of a membrane , and it allows for a comparison of pre- and post-use membrane cleanliness. Ideally, NWP measurement should be performed after every reuse in a lifespan study.
Product yield and purity are often monitored to ensure that product degradation is not induced by repeated use of a membrane. For cases when the filtration step is used for clearance of host cell and process-related impurities, it is important to monitor clearance at appropriate intervals during reuse studies. Such monitoring can help ascertain whether the efficacy of the clearance step is undermined by reuse.
Figure 1. Data from membrane reuse studies for prospective validation: A) normalized water permeability; B) step yield; C) endotoxin levels; and D) conductivity at retentate end
Examining filter integrity can identify problems such as macroscopic holes in the membrane, cracks in the seals, or improperly seated modules, which can lead to product leakage and loss, or unsatisfactory clearance of impurities. A variety of methods like air diffusion tests are available for this purpose.
Transmembrane pressure (TMP) is the force by which the liquid moves through the membrane. The TMP-versus-flux curve often serves as a qualitative indicator of the performance of a membrane step. A carryover of product or impurities often results in decay of the TMP-versus-flux curve. Spectroscopic methods such as Fourier-transform infrared (FTIR) and Raman spectroscopy can be used to analyze the membrane post-use to quantify the buildup or absence of protein.
Performing periodic blank runs at appropriate intervals is a common method used to evaluate cleaning efficacy and potential for product carryover. In a blank run, the membrane step uses load material that does not contain any product, and the resulting pool is analyzed for product-related or other impurities. Analytical techniques often used for this purpose include high-performance liquid chromatography (HPLC) assays, product-specific enzyme-linked immunosorbent assays (ELISA), sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) assays, and total organic carbon (TOC) analysis.
Table 1. Results from the scale-down modeling and qualification
Concurrent validation and prospective validation are two widely accepted approaches for establishing membrane lifetimes. 2,3
Concurrent validation involves performing cycling studies solely at large scale. Carryover and performance are assessed at preset frequencies, and a determination is made on the acceptability of membrane reuse. Lots manufactured in the meantime are quarantined and released once the carryover and performance data have been reviewed and found acceptable. For example, to establish 100 reuses at large scale, a strategy could be to evaluate carryover and performance data for every 10 runs, and to release all 10 lots after the evaluation is complete and successful, including the blank run performed after the tenth lot.
In prospective validation, cycling studies are performed at small scale. Once the target lifetime has been established at small scale, it is considered confirmed and validated at full scale. Although this approach entails substantial work at small scale, it reduces the risk of failure at full scale and the resulting loss of batches that could occur if lifespan were solely determined at large scale. To establish 100 reuses at large scale, a strategy for this approach could be to perform cycling studies at small scale to establish 120 reuses, with evaluation of carryover and performance every 20 runs. Once the small-scale data are in place, monitoring could be performed during manufacturing, with evaluation of carryover and performance performed every 20 runs.
Figure 2. Data from membrane reuse studies for prospective validation: A) data from reuse studies; B) data from blank runs for carryover studies; C) flux; and D) normalized water permeability
The following two case studies discuss examples of processes during which these two approaches have been applied successfully.
This case study for concurrent validation presents the performance of a UF/DF step in a process for a high-volume monoclonal antibody product. The membrane under consideration had been successfully used more than 100 times for a similar step in similar process. Hence, a decision was made to undertake concurrent validation with a target of 100 reuses.
Three criteria—step yield, conductivity, and pH of the product retentate pool—were used to demonstrate acceptable process performance of the membranes throughout their lifetimes. These parameters were monitored for every lot. Cleanability, defined as acceptable lot-to-lot protein carryover and microbial control, was determined based on NWP, bioburden, endotoxin, and protein determination for every lot. Protein carryover was assessed by collecting retentate samples during membrane equilibration, and by testing with specific enzyme-linked immunosorbent assays.
Figure 1 presents data from the reuse study. All performance parameters were within the predetermined acceptance criteria range throughout the membrane lifetime study. Membrane performance with respect to step yield (Figure 1B) and conductivity (Figure 1D) was found to be acceptable. NWP (Figure 1A) and endotoxin levels (Figure 1C) illustrate the effective cleaning of the membrane. Bioburden remained below detection limits throughout the study (data not shown). Further, periodic blank runs were performed and the results were found to be within acceptance criteria (data not shown). The collective set of acceptable performance parameters for the UF/DF membrane system showed that the procedures used to operate, clean, store, and maintain the systems provide effective and controlled performance for at least 100 uses.
This case study for prospective validation presents the performance of a UF/DF step in a process for a low-volume microbial product. The UF/DF step occurred early in the process. Hence, there was concern about the effectiveness of the cleaning procedure. To support a 10-lot campaign at the manufacturing scale, 16 runs were performed at small scale.
The first step with this approach was to establish and qualify the scale-down model.3 Some parameters commonly used to evaluate performance of the scale-down model include NWP, product yield, purity, impurity clearance, flux-versus-time curves, and product carryover. It is important to choose an appropriate cleaning method for both efficacy and for membrane performance with respect to leachables and extractables. Feed material must be representative of what will be produced at manufacturing scale.
Scale down is typically performed at operating conditions identical to those at manufacturing scale with respect to parameters such as product loading per unit membrane area, buffer components and properties (e.g., pH and ionic strength), temperature, TMP, and crossflow rate. Finally, the scale-down model should be compared with the equipment and conditions used at manufacturing scale. Although performance indicators like step recovery can differ at the two scales, depending on equipment setup, it is important to understand what the differences are and to treat the small-scale data appropriately when applying the resulting conclusions to manufacturing scale.
In this case study, product pool pH, conductivity, impurity percentage, and recovery percentage were used to assess step performance. As Table 1 shows, the performance of the scale-down model is equivalent to that at large scale with respect to the parameters given, and the performance of the step is comparable across scales. Recovery was primarily used as a consistency indicator and was greater than 100% because different assays were used to measure product concentration in the load and pool samples. Small-scale reuse studies were performed using this scale-down model. NWP and flux-versus-time curves were monitored to assess the cleanability of the membrane. Finally, blank runs were performed after every five runs to assess carryover at small scale and host cell proteins (HCP), and product concentration via ELISA and DNA were monitored.
Figure 2 shows data from the small-scale reuse study. The membrane performance was acceptable based on pool pH, conductivity, impurity percentage, and recovery percentage (Figure 2A). Flux-versus-time curves (Figure 2C) overlapped, and NWP (Figure 2D) was stable over the number of membrane reuses. Carryover for the blank runs (2, 7, 12, 15, and 17) was minimal with respect to HCP, DNA, and product concentration (Figure 2B). Higher values for host cell proteins and ELISA samples for run 15 were attributed to sampling error, as supported by results for run 17. The small-scale data presented here suggested that 10 reuses could be validated at commercial scale, and this was eventually proven.
The application should dictate the approach used to establish membrane lifespan. Substantial time and resources are required to create and qualify a scale-down model and perform small-scale reuse studies, but when a strong likelihood of cleaning failure exists, these studies are justifiable. Robust routine in-process testing may be substituted for reuse validation, and it may provide a strong assurance of membrane performance. Because lifetime studies for membrane reuse require sampling and testing between lots, the studies may have an impact—sometimes a significant one—on production schedules. A cost–benefit analysis can often substantiate the benefits of validation.
Anurag S. Rathore, PhD, is director of process development at Amgen, Inc., Thousand Oaks, CA, 805.447.1000, firstname.lastname@example.orgR. Samavedam, T. Kichefski, and S. Cote are all senior validation engineers at Amgen, Inc., West Greenwich, RI. R. Morrison is a manager at Commissioning Agents, Inc., Stonington, CT.
1. Rathore AS, Karpen M. Economic analysis as a tool for process development: Harvest of a high cell density fermentation. BioPharm Int. 2006 Nov;19(11)56–63.
2. Parenteral Drug Association. PDA technical report 42: Process validation of protein manufacturing. PDA J Pharm Sci Technol. 2005 Sep–Oct;59 Suppl 4:1–28.
3. Rathore AS, Sofer G. Life span studies for chromatography and filtration media. In: Rathore AS, Sofer G, editors. Process validation in manufacturing of biopharmaceuticals: Guidelines, current practices, and industrial case studies. 2nd ed. Boca Raton: CRC Press; 2005:169–203.