Trends in CCIT Part I: Probabilistic to Deterministic

In the past four or so years working with clients to develop and validate container closure integrity (CCI) test methods, there have been a number overarching trends that have become readily apparent. This blog is the first in a series of three aimed at identifying and discussing such trends.

The first trend comes in the form of a marked shift from probabilistic test methodology to the employment of quantitative, deterministic test methods for use in assessing CCI. A probabilistic test method relies on a series of events, each with a respective probability distribution, to generate results.  Thus, the generation of an accurate result is left to the laws of probability.

Using the example of microbial ingress, in which a sample is submerged in a media solution latent with bacteria, also called a “bug bath”, one can begin to understand the shortcomings of a probabilistic method. In order for a sample to be rejected under a typical microbial ingress regimen, bacterial growth must be present within the sample at a time point post-test. While this seems rational enough, it is critical to understand the sequence of events that must occur, consequently, in order for a leaking package to be identified.

For example’s sake, consider a standard 10 mL glass vial sealed with an elastomeric stopper and aluminum crimp seal, which exhibits a defect in the form of a 5 micron in diameter pinhole through the vial wall. Detection of this defect is subject to the probability of a series of events. Imagine the defective sample is currently submerged in a “bug bath”. In order for detection to occur, bacteria must first be present at the precise location of the defect relative to the entire package. If this does in fact occur, the next required step becomes apparent: the microbe must migrate or proliferate through the defect path. Often this requires that liquid also be present through the length of the defect path. If the organism does make it through the defect and into the package, it now must be able to grow within the package at a sufficient rate to ensure detection during post-test analysis, be it by visual inspection or other means. To use an idiom whose origin lies on a much larger scale, the “stars must align” in order for the defective package to be identified.

Upon contemplating the difficulty in detecting a defective package using the probabilistic microbial ingress test, one may begin to wonder: “If microbial contamination is what we are concerned with, and it is sufficiently difficult to promote microbial ingress even under ideal test conditions, which would never be experienced in the field, why do we bother testing at all?”.

While there may be some debatable truth to this question – it is far less likely that the same package would allow for microbial ingress say, sitting on a pharmacy shelf – the risk for patient safety is still there, no matter how small. Additionally, microbes are not the only element that, if introduced, may have negative effects on the product. Many drug products may be impacted by reactive gases, such as oxygen, which will diffuse through the defect over time, even in the absence of a driving force, such as a pressure differential.  Thus, for both patient safety and product quality, evaluation of CCI should be performed. What, then, should be employed to assess CCI?

The answer to that question is quantitative data generated using a scientifically sound and deterministic test method. In contrast to a probabilistic method, a deterministic method relies on a predictable chain of events driven by known physicochemical phenomena. For example, in a helium tracer gas test, in which a sample may be flushed with helium and subjected to absolute vacuum, if there is a defect through which the helium can theoretically pass, it will pass through as a result of a 1 atm pressure differential between the two sides of the defect. Compare this to the microbial ingress method, in which we know that even if there is a defect through which a microbe may theoretically pass, there are confounding variables that affect the likelihood of such an occurrence. Thus, the helium tracer gas test is deterministic in nature: there is no element of randomness to the results.

Fortunately, in the past four years, companies have become increasingly aware of the stark difference between probabilistic and deterministic methods. Thus, deterministic methods have become all but the standard. Whereas in the past Whitehouse Labs would get consistent requests to develop and/or validate a dye or microbial ingress test from clients with little to no awareness of deterministic methods, inquiries come in a new form these days. With increasing frequency, Whitehouse Labs receives calls from people who have heard about the downfalls of probabilistic methods, or have received pushback from a regulatory agency regarding the use of a probabilistic method, and are now interested in learning more about deterministic methods and how to implement them.

Guidance is beginning to catch up to the industry shift from probabilistic methods to deterministic methods. The most recently proposed revisions to USP chapter <1207> exemplify this. The outgoing, albeit currently official, version of USP <1207> is comprised of 3 pages worth of material “address

[ing] the maintenance of the microbiological integrity of sterile product packaging”. In contrast, the proposed revision, currently available in the USP PF, contains four subsections totaling around 50 pages of material with a clear focus on deterministic methods.

If guidance documents and regulatory agencies continue to move toward deterministic and quantitative CCI analyses, what is now considered an industry trend now may soon become an industry requirement.

Brandon Zurawlow
Associate Director, CCIT