The role of sterile filtration in the biopharm industry

January 7, 2011
Maik W. Jornitz
Pharmaceutical Technology Europe

Volume 23, Issue 1

Biopharmaceuticals commonly cannot be terminally sterilised, as such aseptic processing using sterilising grade filtration is essential.

Biopharmaceuticals commonly cannot be terminally sterilised, as such aseptic processing using sterilising grade filtration is essential. Heat sterilisation, or any energetic input into biotech drug products, causes undesired product degradation so the only available option is sterilised filtration of the biopharmaceutical fluid. Sterilising membrane filtration has been used for decades, and there is a very high experience and expertise level within the industry. As such, the process is reliable and well understood when properly validated. Filtration is also often required to reduce bioburden levels within process streams to avoid potential biofilm formation; for example within purification steps and/or elevated endotoxin levels.

Getty/ Tom Nule

Process validation of sterilising grade filters is a fundamental task to ensure and verify the performance of the filter used within the process conditions and fluid stream. Process validation means that the specified membrane filter is subjected to a bacteria challenge test in accordance with ASTM 838-05, utilising the process parameters of the filtration process and, if possible, the actual fluid product. If the fluid has properties that are detrimental to the challenge organisms, a placebo as close as possible to the actual drug product, is required. The effect of the fluid stream on the organisms needs to be tested using viability tests.

An additional validation test is chemical compatibility to check whether or not the filter membrane and construction is suited for both the product to be filtered and the process conditions. Leachable and particulate matter tests may also be necessary to determine any foreign subject matter release into the drug product. Often forgotten, but essential for biopharmaceutical filtration, are unspecific adsorption studies, which not only determine the fouling rates of the membrane polymer, but also the yield losses onto the filter membrane. These tests can be performed when filters are tested and scaled in filterability trial work. When filters are in line steam sterilised, the sterilisation process requires qualification to prove that it does not damage the filter. Other qualification data, such as gamma sterilisation robustness and efficiency, endotoxin level, integrity test limits and oxidisable substances can be obtained from the filter manufacturer.

The author says...

Over-sterilisation

Over-sterilisation is sometimes cited as a concern. Beside other key attributes, sterility is key to an injectable drug product; therefore I would hesitate to utilise the word "over-sterilisation". Safety and sterilisation robustness are essential, and one has to ensure that sterility is achieved by appropriate measurements, qualification and validation. If one wants to classify something as "over sterilised", it could be through the use of either redundant filtration or filter ratings, which are inappropriate for the application.

More important than adding filters or tighter pore sizes are process validation procedures, which verify that the filter does what it is supposed to do: retain the organism level and species found in the upstream fluid and create a sterile effluent. When a sterilising grade filter is properly validated and evidence has been obtained that it performs to required specifications, why would one need a second one or move to a tighter pore size?

The only reason would be insurance, which creates an immediate insecurity notion. Another "over-sterilisation" aspect could be the European restriction of a maximum 10 cfu/100 mL in front of a sterilising grade filter. Why are end-users of sterilising grade filter bound to such a restriction, when the filters are commonly challenged with 107 organisms per cm2? If one utilises the example of a 1000 L volume with a bioburden of 10 cfu/100 mL being filtered through a 10 in. filter element of 0.6 m2, the challenge would be 16 organisms per cm2, far below the validation challenge test level of 107 per cm2. That sounds very much like overkill!

Regulator concerns

The main concerns of regulators when it comes to sterile filtration depend on which regulator is consulted. Critical regulators very much stress the benefits of heat sterilisation versus sterilised filtration, a known fact acknowledged by everyone who has any scientific sense. Having said that, it would be an advantage if regulators, in contrast, acknowledged the fact that there are drug products that cannot be heat sterilised; in these instances, filtration is the best option.

In addition, when will regulators acknowledge the thousands of process validations performed by the industry with sterilising grade filters, the billions of litres filtered and the decades of experience? If similar scepticism were attached to flying, planes probably would only be used for short distance flights. The regulatory promotion of redundant filtration, pre-use integrity testing or the use of a tighter pore size should be left to the end user, as the decision to use either or all of the aforementioned is an economic decision after the filtration process has been properly validated.

The requested process validation (by all regulators) of sterilising grade filters is necessary and welcomed by filter manufacturers and end users alike. The stated tasks required during process validation are of help to any validation exercise and create a robust performance analysis of a sterilising grade filter. However, it would be good if regulators relied on their stated requirements when the industry performs the requested tasks and documents the outcome and reliability.

Optimising and improving current processes

The sterilising filtration of biopharmaceuticals generally means that a high value, low volume fluid is filtered. As such, oversized filtration systems, redundant filtration systems and, in single-use instances, tubing length, all contribute to excessive hold-up volume, i.e., valuable product losses. Therefore, the size and design of filter systems should not be neglected because these can impact the production process.

Again, unspecific adsorption, which is often neglected, should be tested very early in the drug development process to understand which filter membrane polymer is optimal for a particular fluid or application. Yields are very much influenced by unspecific adsorption and can vary greatly between membrane polymers. Total throughput — the volume that can be filtered through a specified filter device — also has an influence on yield. In some application, yield can drop dramatically when the filter starts blocking. This drop-off has to be determined in small-scale studies utilising flat disc filter composites — so-called indicator trials. When a filter needs to be scaled to production requirements, however, indicator trials are not sufficient and verification trials with small scale pleated devices should be used. The data of these trials are commonly reliable to scale the filter appropriately; again, the goal is not only to find a filter system that is able to process the entire volume, but also not to over-size. High flow rate is the desired parameter in some applications, but flow rate cannot be tested with 47 mm disc filters because the results would be meaningless. The flow dynamics and the filtration area within a 10 in. element requires testing in high flow applications. The interaction between the filter adapter and housing recess, as well as pipe work, the filter pleat behaviour under pressure and possible compression, can only be tested when the actual filter element is used.

In essence, a filter system can be compared to a well-defined recipe; too much as well as not enough can spoil it. Similarly, too much filtration area is as undesirable as too little.

Maik W. Jornitz is Senior Vice President, Marketing, at Sartorius-Stedim North America, Inc. maik.jornitz@sartorius-stedim.com