VALIDATION OF BACTERIAL RETENTION BY MEMBRANE FILTRATION: A PROPOSED APPROACH FOR DETERMINING STERILITY ASSURANCE
Sterilization methods are typically studied to prove their reliability and predictability. Reliability implies definition of the principle variables affecting sterilization. Predictability suggests the ability to forecast assurance of sterility. Filtration, using membrane filters, is the least well characterized sterilization method with respect to reliability and predictability.
Pseudomonas diminuta ATCC 19146, selected as a biological indicator, was optimized for its resistance to removal by filtration. Filtration equipment was designed to allow systematic study of experimental variables. Test methods were developed to allow quantitation of bacterial retention and to define the effects of several parameters (e.g., bacterial numbers, filtration pressure, time and fluid chemistry) on retention. These methods were also used to measure the prediction of retention by a physical test of membrane filters.
The impact of experimental variables studied was a function of filter pore size. Specifically, bacterial retention by membrane filters commonly used for sterilization was independent of bacterial numbers, filtration pressure and fluid chemistry and dependent on time. Sterility of filter effluents was consistently achieved by a single layer of a 0.22 (mu)m pore size filter for continuous filtrations up to 16 hours long.
The bubble point of membrane filters was a strong predictor of bacterial retention. No bacterial passage was observed above a minimum bubble point value. Microscopic examination of the penetration of bacterial cells as a function of depth within a filter suggested a model of bacterial retention based on a sequential sieving of cells by the three dimensional structure of a filter. The knowledge of filter removal ability was combined with the volume and microbial content of liquids to calculate the probability of sterility assurance.
Sterilization by filtration was found to be a reliable and predictable method when applied under properly controlled conditions.