The Bugs Don't Lie

The Belief

The December 2017 draft revision to the European Medicines Agency (EMA) Annex 1: Manufacture of Sterile Medicinal Products reasserts a long-standing position on moist heat sterilization processes that is diametrically opposed to that held by many of the world's sterilization scientists (1): 8.51—Chemical or biological indicators may also be used, but should not take the place of physical measurements.

Identical wording can be found in earlier regulations from the United Kingdom's Medicines Control Agency (now called the Medicines and Healthcare Products Regulatory Agency or MHRA) (2). That the biological indicator is considered a nonessential aspect of the steam sterilization cycle effectiveness contradicts numerous regulatory expectations outside the European Union (3, 4). This publication will review sterilization science to establish whether the EMA stated regulatory expectation is founded in science or is merely opinion-based.

The Premise

Inherent in the EMA's statement is the belief that demonstration of certain physical attributes during a moist heat sterilization process is sufficient evidence that the materials in question have been sterilized. The unstated but implied condition is demonstration of 121°C for 15 min throughout the load as provided in the European Pharmacopoeia (Ph. Eur.) (5). The belief is that demonstration of this physical condition is sufficient to assure effective sterilization.

The Error

The fundamental problem with the preference for physical measurements is the misdirection that it causes. As sterilization processes are intended for the destruction of microorganisms, placing full reliance on physical values implies that they are somehow direct evidence of microbial destruction. That defies reason: the physical measurements must have some original basis in microbial death. The Ph. Eur. stated conditions are likely founded upon some long-lost connection between the stated condition and the reliable destruction of all microorganisms present. Similar statements regarding the secondary role of biological indicators and the reliance on identical sterilizing conditions can be found in the 1983 version of the MHRA Orange Guide (6). The origin of these is equally obscure. The original rationale being lost, it is nevertheless useful to consider the possible evolution of the expected physical conditions.

The Science

Sterilization processes began to evolve in the late 1700s and the first steam sterilizer was built by Charles Chamberland in 1880 (7). Harriette Chick in the early 1900s was the first to describe the logarithmic order of death for microorganisms, which provides a foundation for modern sterilization science (8). The rate of death in moist heat sterilization is linked to the temperature, with kill occurring more rapidly as the temperature increases. The inverse slope of the logarithmic death curve is the D-value, which is dependent upon a number of factors (e.g., microorganism, substrate, and medium). D-values are commonly determined by fraction-negative or survivor curve studies in which the experimental data allows for calculation of the D-value. The D-value provides a means for estimation of safety levels associated with sterilization processes by determination of the probability of a non-sterile unit (PNSU) (see eq 1).1

Where PNSU is the logarithmic probability of a non-sterile unit (sterility assurance level or SAL), F is the delivered lethality (equivalent time at reference temperature), D is the D-value of the microorganism, and N₀ is the initial population of the microorganism.

A minimum PNSU value is used to provide an added measure of safety to the calculation and additional safety factors can include assuming a high population and extreme resistance of the bioburden present. The F value determined corresponds to the assumed physical conditions of the desired sterilization process. The number of estimations, approximations, and assumptions in the arithmetic determination of the F value for the process are substantial (9) as follows:

1. Microbial populations are never precisely known;

2. Microbial resistance is not constant;

3. Growth media varies in its ability to support growth;

4. Microbial death curves aren't always straight lines; and

5. Estimation of the death curve slope is non-exact.

One indication of the inherent variability is the commonly expected tolerance surrounding D value determination of ±20% in the same laboratory (10). Real-world variations in resistance to sterilization are understood to be even greater when different laboratories, different microorganisms, varying substrates, and media variation are all present.

The Ph. Eur. stated norm of 121°C for 15 min throughout the load is equivalent to delivering an F₀ of 15 min at all locations. This F value can be inserted into eq 1; however, the other variables (bioburden population and moist heat resistance) necessary for computation of a PNSU are not provided in the Ph. Eur. content, so the expected level of sterility assurance is indeterminate! Although we have a defined physical expectation, there is no means to assess its appropriateness for the destruction of a specific microorganism. In order to determine the minimum PNSU, assumptions regarding microbial population and resistance have to be made. Thus, attaining a standard physical condition establishes that and almost nothing else about the robustness of the actual sterilization process. Confidence in the robustness of a sterilization process meeting only the Ph. Eur. standard cycle expectation is thus inappropriate.2 To determine the PNSU for a moist heat sterilization process, the microbiological population and resistance of the target microorganisms must be known.

Sterilization Process Realities

The individual microorganisms present in a system will die only when they are directly exposed to lethal conditions; however, what those exact conditions are at the point of kill is unknown because physical measurements only describe a portion of what is necessary to establish sterilization efficacy. Microbial death in a moist heat system requires that the microorganism be simultaneously directly exposed to heat and moisture. One could argue that because physical measurements are inherently conservative (especially when safety factors are incorporated into their selection) that validation efforts need only confirm them and that no microbiological evidence is necessary. If only sterilization were that simple. Many factors relating to microbial destruction preclude placing sole reliance on physical measurements for sterilization process validation.

Measurement Error

Temperature measurement can require penetration of the wrapping materials to position the probes within the load items. Unless managed correctly, the means for probe access can compromise the integrity of the wrapping materials and substantially alter the physical conditions observed. Compromised wrapping when measuring temperatures precludes meaningful results. Similar effects can occur when introducing temperature probes into filled product containers. In these instances, physical measurements in improperly probed load items can differ markedly from those in load units where package/container integrity is maintained. The introduction of temperature probes into prefilled syringes, ampoules, and small vials can alter the container such that the suitability of the results must be questioned (see Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Small vial with thermocouple.

As biological challenges can be inserted and recovered without altering the integrity of the monitored item, the measurement errors inherent in temperature monitoring are avoided.

Mistaken Belief in Mathematical Precision

Although not stated, the Ph. Eur. standard cycle conditions are predicated on historical evidence of microbial destruction at similar conditions. This is derived from the work of Chick described previously, which must be understood as a mathematical approximation of the microbial results. The expected cycle incorporates unstated simplifying assumptions about the sterilization process and the undefined microbiological target. Real-world experience with microbiological challenges in sterilization demonstrates that the precepts of a constant rate of kill (D-value), temperature dependence of the kill rate (z-value), and microbiological population (N₀) vary substantially. This is certain for biological indicators where the established variation in resistance (which requires knowledge of the parameters mentioned above) is approximately ±20% in the same laboratory (10). Variations with wild-type microorganisms should be understood as even greater. Nevertheless, direct evidence from microbiological results is always preferable to the indirect thermal data.

Nonlinearity of Microbial Death Curves:

As previously stated, the use of physical measurements for proof of sterilization is linked to assumptions regarding the microbial resistance as established through D- and z-value determination. This is often represented by a linear death curve as depicted in Figure 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Typical microbial death curve.

Representing the death curve as a perfectly straight line, although prevalent in most publications, is a simplification of microbiological realities. There is documented evidence of death curves that are decidedly nonlinear including such variants as concave up, concave down, and so forth (9). The existence of these anomalous death curves weakens the argument for physical data correlation with microbiological evidence. The outcome of a microbiologically challenged validation study where the death curve is nonlinear would follow the actual death curve and not the mathematical prediction (see Figure 3).3 Unfortunately, none of this can be confirmed experimentally as the slope of the death curve is unknown beyond the point where viable microorganisms can no longer be recovered. The mathematical estimation of outcome is less reliable than real-world microbiological evidence under those circumstances.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Extrapolated death curves.

There are other objections regarding the use of biological indicators unrelated to the uncertainties of physical measurements. The more commonly indicated concerns are detailed in the following sections.

Summation

The physical conditions associated with effective sterilization (that is, overkill) are stated in the Ph. Eur. or PDA's Technical Report #1, Validation of Moist Heat Sterilization Processes: Cycle Design, Development, Qualification and Ongoing Control.

For this method of terminal sterilisation the reference conditions for aqueous preparations are heating at a minimum of 121°C for 15 min. (5)

A sterilization design approach where minimal information is required about the product bioburden. A worst-case bioburden assumption is used to determine the delivered lethality needed to achieve a PNSU of 10⁻⁶ on or in the items being sterilized. When using this approach, the qualification program must demonstrate that both the F_BIO and F_PHY are greater than 12 minutes (19).

The referenced documents, either indirectly or explicitly, include expectations for SAL and PNSU that are associated with microbial destruction. Thus, although it might seem that the physical conditions sufficiently describe a satisfactory sterilization cycle, the stated processes rely on unstated assumptions related to microbial destruction. The cited values only have validity because substantial microbiological evidence had been previously gathered to support their efficacy. Additionally, although it may appear that physical measurements are more definitive and precise, they are little more than convenient placeholders based upon an assumed PNSU and expected bioburden population/resistance. To claim that physical measurements should take precedence in sterilization validation is clearly unfounded when the origin of those physical measurements being utilized is biological destruction. Physical methods are an expedient means of assessing cycle performance, but they should never be considered definitive evidence.

As a young man, the author was a member of a competitive rifle team. In order to establish one's score the targets were retrieved and examined to confirm the results. Assumptions based upon rifle position, sight accuracy, trigger actuation, bullet weight, and speed could never replace the direct evidence obtained from examination of the target itself. Sterilization is best assessed in the same fashion.

Sterilization is a critical part of processes for essential therapies and its successful operation is vital to patients worldwide. Assuring the safety of sterile products requires that sterilization processes be appropriately developed, validated, and executed. Sterilization processes rely on equipment and systems that deliver conditions lethal to microorganisms. To ensure their reliability a combination of both physical and biological knowledge is essential. Of those, microbiological evidence must be recognized as definitive, with the temperature data understood as merely corroborative.

Note

This article expands upon earlier publications that addressed the relationship between microbiological and physical methods for the evaluation of steam sterilization effectiveness (20, 21). It was prepared because in the nearly two decades that have elapsed since the first publication, the misinformation embodied in the EMA's preference for physical measurements persists. With EMA Annex 1 currently under revision, the potential to correct this major flaw in its sterilization recommendations exists. The reader is encouraged to comment upon this and other aspects in the revision draft of Annex 1.

Conflict of Interest Declaration

The author declares that he has no competing interests.