Laboratory replication of filtration procedures associated with Serratia marcescens bloodstream infections in patients receiving compounded amino acid solutions

Purpose. Specific deviations from United States Pharmacopeia standards were analyzed to investigate the factors allowing an outbreak of Serratia marcescens bloodstream infections in patients receiving compounded amino acid solutions. Methods. Filter challenge experiments using the outbreak strain of S. marcescens were compared with those that used the filter challenge organism recommended by ASTM International (Brevundimonas diminuta ATCC 19162) to determine the frequency and degree of organism breakthrough. Disk and capsule filters (0.22- and 0.2-mu m nominal pore size, respectively) were challenged with either the outbreak strain of S. marcescens or B. diminuta ATCC 19162. The following variables were compared: culture conditions in which organisms were grown overnight or cultured in sterile water (starved), solution type (15% amino acid solution or sterile water), and filtration with or without a 0.5-mu m prefilter. Results. Small-scale, syringe-driven, disk-filtration experiments of starved bacterial cultures indicated that approximately 1 in every 1,000 starved S. marcescens cells (0.12%) was able to pass through a 0.22-mu m nominal pore-size filter, and about 1 in every 1,000,000 cells was able to pass through a 0.1-mu m nominal pore-size filter. No passage of the B. diminuta ATCC 19162 cells was observed with either filter. In full-scale experiments, breakthrough was observed only when 0.2-mu m capsule filters were challenged with starved S. marcescens in 15% amino acid solution without a 0.5-mu m prefiltration step. Conclusion. Laboratory simulation testing revealed that under certain conditions, bacteria can pass through 0.22- and 0.2-mu m filters intended for sterilization of an amino acid solution. Bacteria did not pass through 0.2-mu m filters when a 0.5-mu m prefilter was used.