Pharmaceutical container/closure integrity. II: The relationship between microbial ingress and helium leak rates in rubber-stoppered glass vials.

Helium leak rate measurements were quantitatively correlated to the probability of microbial ingress for rubber-stoppered glass vials subjected to immersion challenge. Standard 10-mL tubing glass vials were modified by inserting micropipettes of various sizes (0.1 to 10 microns nominal diameter) into a side wall hole and securing them with epoxy. Butyl rubber closures and aluminum crimps were used to seal the vials. The test units were sealed in a helium-filled glove bag, then the absolute helium leak rates were determined. The test units were disassembled, filled with media, resealed, and autoclaved. The test units were thermally treated to eliminate airlocks within the micropipette lumen and establish a liquid path between microbial challenge media and the test units' contents. Microbial challenge was performed by immersing the test units in a 35 degrees C bath containing magnesium ion and 8 to 10 logs of viable P. diminuta and E. coli for 24 hours. The test units were then incubated at 35 degrees C for an additional 13 days. Microbial ingress was detected by turbidity and plating on blood agar. The elimination of airlocks was confirmed by the presence of magnesium ions in the vial contents by atomic absorption spectrometry. A total of 288 vials were subjected to microbial challenge testing. Those test units whose contents failed to show detectable magnesium ions were eliminated from further analysis. At large leak rates, the probability of microbial ingress approached 100% and at very low leak rates microbial ingress rates were 0%. A dramatic increase in microbial failure occurred in the leak rate region 10(-4.5) to 10(-3) std cc/sec, which roughly corresponded to leak diameters ranging from 0.4 to 2 microns. Below a leak rate of 10(-4.5) std cc/sec the microbial failure rate was < 10%. The critical leak rate in our studies, i.e. the value below which microbial ingress cannot occur because the leak is too small, was observed to be between 10(-5) and 10(-5.8) std cc/sec, which corresponds to an approximate leak diameter of 0.2-0.3 micron.