Choosing and using biological indicators for moist-heat and VHP processes

A biological indicator is supposed to be the instrument of truth in a sterilization or decontamination process — a calibrated population of resistant spores whose survival or death tells you, unambiguously, whether the process achieved its lethality. But an indicator is only truthful if it genuinely represents the challenge the process must overcome. Choose the wrong one, place it carelessly, store it improperly, or assume its certified resistance transfers to your equipment, and the indicator will give you a confident answer that happens to be wrong. The discipline of biological indicators is, at its core, the discipline of making sure the instrument is not lying to you.

What a biological indicator actually is

A biological indicator (BI) is a defined population of a resistant microorganism on a carrier, characterized by three linked properties: a specific population, a specific D-value, and specificity to the microorganism, the substrate, and the process technology. The D-value is the time, in minutes, required to reduce the population by one log (90%) under defined conditions. For moist-heat sterilization the organism is typically Geobacillus stearothermophilus; for vaporized hydrogen peroxide (VHP) decontamination it is G. stearothermophilus on a stainless-steel substrate. The carrier and process matter as much as the organism: the same spore population behaves differently on different surfaces and under different sterilants, which is precisely why a BI is specified for a process and a load, not as a universal token of sterility.

Moist heat: when the load itself changes the answer

Consider a common aseptic-filling load — rubber stoppers, which must be sterile before transfer into an isolator. Stoppers are not heat-sensitive, so moist-heat sterilization (a saturated-steam cycle in an autoclave, whether in bags or in bulk) is the natural choice, and the overkill approach applies: because the items are essentially unaffected by the process, destroying a high concentration of a resistant organism demonstrates destruction of any reasonably anticipated routine bioburden, giving a high margin of safety with welcome simplicity.

The subtlety lies in how the substrate alters spore resistance. The European Pharmacopoeia (11.2, 5.1.2) is explicit that the load and product should be assessed to find the most difficult position to sterilize — cold spots, the vial–stopper interface, hard-to-penetrate areas — and that the biological challenge should simulate that worst-case position as closely as possible. Critically, it warns that spores inoculated into a product or onto surfaces are known to react differently to sterilizing conditions compared with biological indicator units , and that in such cases commercially available BIs may not be suitable; an inoculated test item prepared from a well-characterized spore suspension may be the better model.

This is not a theoretical caution. When G. stearothermophilus is inoculated directly onto the surface of a 13 mm stopper and its D-value determined at the cycle temperature, the result can come out higher than a standard BI's — 3.6 minutes at 121 °C in one case — meaning the inoculated stopper is a more demanding, more representative challenge, and can be used to validate the cycle directly. But the relationship is not always in that direction, and the kinetics can be more troubling still.

The tailing effect: when the kill curve bends

A fraction-negative D-value study on stoppers reveals a phenomenon that should give every validation specialist pause. For a conventional, well-behaved BI exposed uniformly, the fraction-negative result is orderly: a constant decrease in positives until total kill, across a short fractional window of a few minutes. Real stoppers do not always cooperate. In a steam-resistometer study — twenty stoppers per exposure, each interval a constant step longer, cultured into tryptic soy broth and read for growth over seven days, with the D-value calculated by the ISO 11138-1-compliant Stumbo–Murphy–Cochran fraction-negative method — the data showed a pronounced tailing effect : spores were easily killed at first, but as exposure increased the survivors became progressively harder to kill. The kill curve was not linear.

The computed D₁₂₁ of 1.9 minutes was, in this situation, a misleading average. It does not represent the real behaviour of spores on that stopper surface — likely shorter at the beginning and longer later — because the inactivation kinetics are non-linear and afflicted by tailing. And here is the trap: most commercial BIs have a D₁₂₁ greater than 1.9 minutes and a clean linear kill curve untroubled by tailing. A naïve comparison would suggest the BI is the more conservative (more resistant) challenge and therefore the safe choice. It is not. Because the BI cannot reproduce the inoculated stopper's tailing kinetics, it will not represent the cycle's true killing behaviour. The correct response is to abandon the predictive BI comparison and adopt an empirical approach : run cycles, progressively increase the exposure phase until total kill, then double it to assure a SAL ≤ 10⁻⁶. The indicator that looked safer would have lied.

A practical end-user procedure follows from this. Stoppers with autoclave tape are bagged individually; roughly two hundred are sent to a validation laboratory at least a month ahead; the laboratory inoculates them, performs the D-value studies, and certifies identity, count, purity, and D-value; inoculated stoppers and a vial of the same inoculum batch are shipped to the manufacturing site, which independently tests both on arrival for identity, count, and purity, using the certified D-value for cycle validation.

VHP: a world without a standardized D-value

If moist heat demands care, vaporized hydrogen peroxide demands a different mindset entirely — because for VHP, a standardized D-value cannot exist . The reason is intrinsic to the sterilant. Hydrogen peroxide decontamination is biphasic: even though it is described as a "dry" technology, a wet phase is always present, and the kill rates in the gas and liquid phases differ substantially, reflecting different concentrations and available water in each phase (Akers & Agalloco, Pharmaceutical Technology , 2013). The linearity of microbial destruction cannot be assured, because the process is not completely homogeneous.

The field's own framing is unusually candid: in VHP, a standardized equipment, a standardized cycle-management system, a standardized cycle, and a standardized recipe all do not exist — and consequently neither can a standardized BI D-value. USP <1229.11> states the conclusion directly: standard sterilizing conditions have not been defined because of the varying and multiphase nature of the sterilant, no standardized BIs with D-values suitable for conventional predictive kill-rate analysis exist, and in their absence an empirical approach must be used . As one authoritative treatment puts it, the manufacturer's D-value should never be used to estimate process dwell in the user's system; accurate D-value determination requires replicate exposures at known constant conditions, and neither requirement is present in a large-scale vapor system (Agalloco & DeSantis). The practical resolution is to determine a system D-value specific to your own equipment, technology, and recipe — and to read the results of your own cycle to develop the correct recipe. The PDA's guidance reinforces the discipline: the D-value should be known and comparable from one validation to the next so that any shift in VHP efficacy can be detected.

Using BIs correctly: placement, storage, and population

Even a correctly chosen BI fails if it is used carelessly, and VHP is especially unforgiving:

• Placement. Hydrogen peroxide needs space to penetrate, so BIs must not be placed completely flat against a surface — a horizontal, flush position risks failing to decontaminate one side of the BI. They should sit so the sterilant can reach both surfaces and penetrate beneath the Tyvek pouch, located next to or on the surface of interest. The Tyvek pouch is itself a considered choice: hydrophobic, with a high moisture-vapour transmission rate, a good microbiological barrier well suited to vaporized and dry processes — but its hydrophobicity means a liquid hydrogen-peroxide phase may penetrate it less, lowering the dose the spores actually see.
• Storage. BIs are stored at 2–8 °C and below 50% relative humidity. This is not arbitrary: studies show BI resistance varies with relative humidity across different exposure times, and resistance stays consistent below 50% RH — so the storage limit exists to prevent a shift in resistance that would corrupt the cycle results (PDA TR 51).
• Population. ISO 11138 allows confirmed population to fall within 50–300% of nominal; customers diverge on whether they prefer lower populations (to minimize aggregates and false positives) or higher ones (to stay above an internal 1×10⁶ CFU floor), and the BI manufacturer should be consulted so consistent techniques are used.

When a BI comes up positive: the mandatory investigation

A positive BI at the end of a cycle is not, by itself, proof that the process failed — and the investigation is mandatory. It must look in three directions at once: the equipment, the process, and the biological indicator itself. For the BI, the questions are concrete: was it correctly located, correctly chosen for the cycle and pouch, properly stored — and could it be a post-contamination, or a rogue BI , a hyper-resistant outlier? A broader strategy compares results across cycles and investigates trends in hydrogen peroxide concentration, temperature, relative humidity, and injection rate, alongside filling-room conditions, BI storage and handling, load configuration, the physical locations of the positives, calibration and maintenance, and H₂O₂ degradation.

Rogue BIs deserve special attention because they expose a tempting shortcut. Their causes are physical: spores clumped into aggregates, spores coated in debris, and unintended crevices or cracks in the carrier surface. The shortcut is to assume that a scanning electron microscope (SEM) image can predict BI performance — and some firms have begun leaning on this subjective test as a primary incoming-inspection criterion. A controlled experiment with BIs from four manufacturers shows why that is dangerous. The "best-looking" presentation under SEM — a clean spore monolayer — was not reliably the best performer, and a worst-case appearance (spores covered by heavy film) did not reliably predict failure; performance in an actual test isolator was what separated heavy late-positive tailing from clean kills. Worse, isolators are not uniform, so a BI imperfection that is harmless in one system can be problematic in another — which is exactly why one client will reject a lot that another favours. The ideal of a perfect monolayer of identical spores on a flawless surface, all dying in unison, is unattainable; the best screening method is to evaluate a lot's performance in a test isolator operating under the same parameters as production, which assesses the BI in its entirety rather than one photogenic aspect of it. SEM is not discouraged — but one should be cautious about what the image is really telling you.

Conclusion

The recurring lesson across moist-heat and VHP validation is the same: the biological indicator is a tool whose truthfulness must be earned, never assumed. Every process has its specific BI for a specific need and a specific load, and the certified resistance printed on a certificate is a reference point, not a transferable fact — least of all in the multiphase world of vaporized hydrogen peroxide, where no standardized D-value can exist and an empirical, system-specific approach is mandatory. An inoculated stopper with a tailing kill curve, a BI flush against a surface that VHP never fully reaches, a lot judged on its SEM portrait rather than its performance — each is a way for the instrument of truth to mislead. The microbiological performance qualification must be designed by validation experts who understand both the process and the indicator, and when a positive appears, the investigation that follows depends entirely on that same dual knowledge. The indicator does not lie on its own; it lies when we ask it the wrong question.

References