Are Terminal Sterile Filters in Water Systems permissible?

One of the most frequently asked questions in the pharmaceutical environment is why so much effort is made for purification and storage/ distribution of water for pharmaceutical use. Especially microbial contamination - the highest risk - could be avoided by using terminal sterile filters at the points-of-use.

Answering the question whether terminal sterile filters are allowed first requires reviewing the legal requirements and standards as well as the guidelines on the topic water. On some points, the EU GMP Guides lays the specifications for premises and equipment. Closely related to the terminal sterile filters, you can find the following instructions:

• EC GMP 3.38: "Equipment should be installed in such a way as to prevent any risk of error or of contamination."
• EC GMP 3.39: "Production equipment should not present any hazards to the products".
• EC GMP 3.36: "Manufacturing equipment should be designed so that it can be easily and thoroughly cleaned."
• EC GMP Annex 1: "Water treatment plants and distribution systems should be designed, constructed and maintained so as to ensure a reliable source of water of an appropriate quality."

Many guidelines also provide references to terminal filters. Please read here an extract from a Japanese Guideline:

"Sterile drug products produced by aseptic processing (Japan 2006)
As a rule, sterilizing filters should not be placed at water use points since the filters could mask microbiological contamination in the water system. Endotoxins could also be released from dead microorganisms retained in the filters. If the use of filters is unavoidable, the interval of replacement should be based on validation results."

The instructions with regard to terminal filters in this Japanese document are clear: basically, terminal filters shouldn't be used. Indeed, they would provide a false security and can't be - in this application - compared to the sterile filtration in sterile production. Here, the integrity of filters is checked before and after their utilisation for each batch.

There may be exceptions, though. Filters can't be used to mask too high CFU-values. Moreover, one should be able to justify the use of such filters instead of ensuring the quality required trough system design, validation and monitoring. Normally, this shouldn't be easy.