Manufacturers of cartridge filters validate their filters in house, in accordance with a standard methodology described in ASTM F8381 using Brevundimonas diminuta as the challenge organism; in brief, at a pressure of 2 bar, a sterilising grade membrane filter is challenged with bacteria grown in an aqueous buffer. A positive control filter of 0.45µm pore size is used to ensure that the bacteria grown are small enough to penetrate any defect in a 0.2µm filter.
It is important to note that both liquid and gas sterile filters in pharmaceutical processes are required to prevent the breakthrough of bacteria in a liquid challenge. Gas sterile filters are often wetted during the process, either inadvertently with the product, or during steam sterilisation. In both scenarios, a filter that is validated with a liquid challenge will ensure sterility.
Ensuring that filtration will result in a sterile product in a process requires the filter to be tested in the process conditions, ideally with the product or a surrogate. Products such as antibiotics can often be viscous and contain aggressive chemicals, and hence the filter needs to be validated to account for the increased pressure drop (due to viscosity) and the chemical effects of the product on the filter membrane, hardware and seals. Figure 1 illustrates a general outline of the process of product validation. The processes required for filter validation are outlined in the Parenteral Drug Association’s Technical Report 26 Sterilizing Filtration of Liquids2.
Figure 1: Overview of validation of a sterile membrane filter process
Manufacturers of pharmaceuticals will often produce a range of pharmaceuticals that vary little by composition and can be pooled into ‘families’ to reduce the cost of testing. In each family the worst-case scenario is chosen, either by combining a range of ingredients or choosing a single product based on known issues of the compatibility of an ingredient with a particular membrane.
A study is initiated by the filter manufacturer to deduce the correct membrane and seal to use. The process of initial selection varies, but bubblepoint testing of the membrane and checking for discoloration and swelling of seals is required. Techniques such as tensile strength testing following product exposure can yield information not only on loss of strength, but just as importantly, loss of ductility. Porometry techniques can yield information on both a shift in pore size distribution and a loss in permeability due to swelling caused by, for example, a solvent-based product.
An assessment is required to check the bactericidal activity of the product. If the product is found to be bactericidal, then a surrogate has to be chosen that closely resembles the product. Any residual product that kills bacteria in the buffer solution invalidates the final sterile filtration bacterial challenge. A surrogate is often chosen for extractables testing where the ingredients might interfere with analytical techniques.
The microbial challenge in a nutrient-rich formulation might be in excess of the safety margin of a sterilising grade filter
Generally the product is bactericidal, and growth of bacteria in processes upstream of filtration is not a challenge. However, some pharmaceutical formulations can encourage microbial growth before the filtration process, e.g. saline or sugar-based solutions. The microbial challenge in a nutrient-rich formulation might be in excess of the safety margin of a sterilising grade filter. In this case, an upstream membrane filter to reduce bioburden will result in a lower bacterial count upstream of the final sterilising grade filter and thus ensure that the product is rendered sterile. In some cases, both a coarse (0.45µm or 0.65µm) and final sterilising grade 0.2µm filter may be used for better process economics. Validation of such a process requires two filters in series to be tested.
Selection of filters for the validation studies differs between manufacturers; the best practice is to select at least one test filter that is close to the specified bubblepoint. The current regulatory requirements are that measurement of bubblepoint is the key parameter for assessing integrity; this can be error-prone with exaggerated bubblepoints often recorded with automated test equipment or a level of subjectivity when carried out manually. It is advisable to use forward flow/diffusive flow testing in addition to bubblepoint. Pleated filters are preferred to discs, for the reason that the folding process itself can degrade the filtration efficiency3.
Membrane filter cartridges are often compromised in processes due to a loss of ductility
Membrane filter cartridges are often compromised in processes due to a loss of ductility; in such a scenario a validation with a pleated filter is essential because the movement of pleats will result in ‘membrane-cracking’ at the peaks and troughs if compatibility is an issue. Invariably, validation with small-scale pleated filters is more difficult than with discs, but with significant benefits for process safety.
Once the filter has been chosen, the product is circulated through the filters for a time equivalent to that in the actual process, with a flow rate chosen by scaling down from the full-scale process. This ‘conditioning’ process can often be fraught with problems due to issues such as the compatibility of the hardware. For example, an acidic, corrosive product is not uncommon and thus the extractables obtained are often from the corroded hardware rather than from the filter itself.
The final product from circulation through the filters is then assessed for extractables with Fourier transform infrared spectroscopy (FTIR) and non-volatile-residue (NVR) analysis commonly used. Additional techniques such as high performance liquid chromatography mass spectrometry (HPLC-MS) can be useful depending on the nature of the product. Chemical compatibility issues are often linked to extractables; an intact filter membrane will generally yield few extractables. In single-use systems where there is a significant amount of polymeric hardware, the extractables content can be much higher.
Small wall-less bacteria that can penetrate 0.2µm filters have been discovered, but are only considered a risk to the safety of those cell culture processes where antibiotics are not used
A separate set of filters that have been conditioned with the product are then flushed (if the product is bactericidal) to ensure that no residue of the product remains on the filters. The filters are challenged with B.diminuta bacteria suspended in the product or a surrogate buffer. A positive control filter of 0.45µm is used as per the ASTM F838 methodology. The general requirement is that no single colony should result following the challenge for an injectable, sterile product. Where a single colony is found, the process has to be repeated either with a filter of smaller pore size, a dual membrane filter or series filtration.
Small wall-less bacteria that can penetrate 0.2µm filters (e.g. Mycoplasma spp) have been discovered, but are only considered a risk to the safety of those cell culture processes where antibiotics are not used. The validation process described is identical except filters described as 0.1µm or 0.03/0.04µm are used to render the product free of Mycoplasma.
Total sterility is not always a requirement: creams and ointments, for example, are often allowed to contain a certain CFU content/ml or gram; here the critical aspect is that no micro-organisms that can cause skin infections are present following filtration of the final product.
Finally, as novel filters and an array of single-use filtration systems come on to the market, the challenge remains to verify that the filter is fit for the processes in which it is used, and not simply in an R&D laboratory with benign solutions in gentle process conditions.
References
1. ‘Standard Test Method for Determining Bacterial Retention of Membrane Filters Utilized for Liquid Filtration’, ASTM-F838-05, 2013
2. ‘Sterilizing Filtration of Liquids’ Technical Report No.26 PDA Journal of Pharmaceutical Science and Technology, Antonsen, H et al. 62, 2008
3. Pall Corporation: Changes in membrane filter capture efficiency due to folding, L.J. Cummings et. al. 2013
