How to Run an ISO 11135 Ethylene Oxide Sterilization Validation

The most practical question first: what does a validation have to deliver

Ethylene oxide (EO) sterilization validation does not end when you have run one cycle and passed a sterility test. Under the ISO 11135 framework it is a full chain: product definition, process definition, installation qualification (IQ), operational qualification (OQ) and performance qualification (PQ), producing a report that demonstrates that this product, in this packaging, loaded this way, on this sterilizer running this programme, consistently achieves the specified sterility assurance level - and, on that basis, a set of routine release rules. What attracts the most deficiency letters in a filing is rarely the sterility result. It is that the middle sections are incomplete, or were done but never joined into a chain of evidence: bioburden data that no longer reflects volume production, a loading pattern that does not match the line, a process challenge device with no equivalence rationale at all.

For a Chinese filing, EO validation normally maps onto the GB 18279 set of requirements; for CE or FDA routes you cite ISO 11135 directly. The two technical frameworks are closely aligned, and in practice most companies run one protocol and reference both numbers in the report. For parameters and acceptance limits, refer to the current valid version of the standard text.

Three things that must be frozen before validation starts

The first is product definition. A single registration unit often contains several variants differing in material, cavity geometry and how tightly the assembly seals. The product definition has to state which models are included in this validation and on what basis. Cut corners here and the product family grouping later has nothing to stand on.

The second is process definition. The parameter ranges for preconditioning, evacuation, humidification, gas charge, exposure, purge and aeration have to be locked before validation starts, not adjusted as you go. The value of a validation report comes from repeat verification at fixed parameters; if the parameters keep moving, every run completed so far is void.

The third is packaging. EO is a gas process: the agent has to penetrate the pack and reach every part of the product, and residuals have to come back out after the cycle. The packaging system is validated separately under the ISO 11607 series, including the ISO 11607-1 requirements for materials and packaging systems, but the schedules have to mesh: running a sterilization validation before the packaging is frozen means throwing away a round. See sterile packaging validation for that part.

What IQ, OQ and PQ each prove

These three stages are easily written as three documents in the same template, when in fact they answer three different questions: was the equipment installed correctly, does the equipment run accurately, and was the product actually sterilised?

The division between OQ and PQ is worth stressing. OQ is about the equipment itself and is usually run empty or with a simulated load. PQ has to be run with real product, real packaging and a real load. We have seen companies use OQ data as PQ on the grounds that the parameters are identical - identical parameters do not mean identical conditions inside a product cavity, and that kind of submission generally comes straight back.

Half-cycle or bioburden approach: what should decide it

This is the earliest decision in an EO validation and the hardest to reverse. Pick the wrong route and every protocol, sample and schedule downstream has to be redone.

The half-cycle approach works by establishing a routine sterilization cycle and then running it with the gas exposure time halved, to see whether biological indicators are still completely inactivated under those more demanding, less favourable conditions. If they are, the routine cycle has adequate lethality margin. Its advantage is that it does not depend on fine characterisation of the resistance of the product bioburden; it is robust in engineering terms, and it is the default choice for most single-use sterile devices. The bioburden approach instead derives the required sterilization conditions from the actual microbial population level and resistance data of the product. It demands continuity and representativeness in the bioburden data, and what you get back is the ability to push sterilization intensity lower.

The basis for choosing is not which one is more advanced, but whether the following conditions hold together.

A few practical rules of thumb. First, if the product is still in design freeze or the supply chain is unsettled, there is essentially no room for the bioburden approach - the population stability the derivation depends on simply does not exist. Second, the extra investment in the bioburden approach only pays off when sterilization intensity is genuinely the bottleneck on product performance; if the product tolerates the process anyway, lowering intensity buys nothing while the added data maintenance cost runs indefinitely. Third, the two routes are not mutually exclusive: a common arrangement is to complete process validation by the half-cycle approach while using bioburden monitoring for routine release and trend management, and to go back and reassess whether the existing margin is still adequate if monitoring shows a rising trend.

Bioburden determination itself - choice of sampling points, elution and enumeration methods, recovery efficiency validation and result review - is a separate methodology and is not covered here; the corresponding method basis is ISO 11737-1, and there is a dedicated article on bioburden determination in the knowledge base. The only point to carry forward is this: whichever route you take, bioburden is an input to the sterilization validation, and its sampling conditions must line up with volume production conditions, or the whole validation report is built on sand.

Exposure times, temperature and humidity conditions, biological indicator types and acceptance requirements are per the current valid version of the standard text. Products differ enormously, and copying a peer's protocol wholesale is a high-risk move.

Loading pattern and process challenge device: the two places things go wrong

On loading pattern, the tray arrangement, stack height and proportion of chamber volume used during validation must represent what actually happens in production. Two extremes are common in practice. In one, the validation load is sparse while production runs a full chamber, so gas penetration conditions are entirely different. In the other, validation is run at maximum load while production frequently runs small batches, and you end up having to justify the applicability of light loads after the fact. The safe approach is to cover both the upper and lower loading limits during validation and to state in the report the loading range permitted in production, with anything outside that range triggering an assessment.

The role of a process challenge device (PCD) is to carry the biological indicator in place of the least favourable location inside the product, so it has to be harder to sterilise than the product itself, or shown by justification to be equivalent to the internal conditions of the product. Devices with long narrow lumens, multiple cavities or sealed fittings - infusion sets, catheters, valve assemblies - deserve particular care. Taping a biological indicator to the outside of the pouch measures the environment outside the pouch; the actual conditions inside the product cavity are never examined at all.

That judgement calls for engineering analysis of the specific construction rather than a template. There are several physical paths to reason along. First, the length and tortuosity of the gas diffusion path: the longer and more convoluted the lumen, the longer the agent takes to reach the far end. Second, points of cross-sectional contraction: fittings, valve seats and filter membranes create sudden narrowings that act as diffusion bottlenecks, and the region just beyond them is often the least favourable point in the whole device. Third, blind ends and dead volumes: air that evacuation fails to clear lingers in a blind end and forms a cushion that blocks the agent. Fourth, sorption and desorption behaviour of the materials: some polymers absorb a large quantity of agent, which both affects the effective concentration during exposure and determines how difficult aeration will be. Fifth, the lag in humidity penetration: moisture typically reaches deep cavities more slowly than gas during humidification, and lethality falls noticeably when humidity is inadequate. Stack those five together and you can normally identify the single hardest-to-sterilise point on the product, and the PCD design or inoculation location should be aimed at it.

There are generally three ways to justify PCD equivalence: design a physical simulation device with clearly greater diffusion resistance than the product; inoculate the biological indicator directly into the least favourable location of the product to create an inoculated product; or place indicators inside the product and in the PCD in parallel and use the data to show that the PCD is no weaker than the product. Whichever you use, the protocol has to spell out why you believe that location is the least favourable one - that sentence is what reviewers are really reading.

Aeration and residuals: build them into the schedule

Finishing the cycle does not mean the product can be released. Aeration has to bring ethylene oxide and its reaction products below the specified levels, and this part is evaluated under ISO 10993-7 / GB/T 16886.7. Three points matter in practice. Residual samples must come from product that has been through the real aeration process; separately ventilated samples do not count. Sampling has to cover the locations where aeration conditions are relatively unfavourable, such as the centre of a tray or the inner layers of a stack. And aeration itself takes real time, as does the testing, which is exactly why many projects stall at this stage - the time was never put in the plan. For the related items see sterilization and residual testing.

One more caution: acceptable residuals are one of the necessary conditions for release, but they are not a substitute for process validation. We have genuinely seen companies submit a residual report as their sterilization validation documentation, and the only possible outcome is a deficiency letter and a restart.

Product families and equivalence: how to legitimately run fewer rounds

Running a full validation on every variant is hard to justify on cost. The workable approach is to group products into families, run a complete validation on the most challenging representative product in each family, and bring the remaining models in by equivalence justification. The justification usually rests on: the material and its sorption and desorption behaviour with EO, geometry and cavity complexity, packaging format and breathable area, load density, and bioburden level and population characteristics.

The judgement here is clearly an engineering analysis. For example, among two silicone components, the one with thicker walls and deeper blind cavities is normally both harder to sterilise and harder to aerate, so it should be chosen as the representative. Or take a variant where the breathable paper area was reduced for cosmetic reasons - its challenge level within the family may in fact be higher. Note also that sterilization challenge and aeration challenge do not necessarily fall on the same model: the one with the longest lumen is hardest to sterilise, while the one with the thickest walls and greatest material sorption is hardest to aerate. In that situation you either nominate separate representatives for each, or choose one model that is not favourable on either dimension and explain the reasoning. The grouping logic goes into the protocol and leaves a trace in the report, because what a reviewer actually cares about is the sentence explaining why you think this one represents that one.

Requalification and change control

Validation is not a one-off. Major overhaul or replacement of key sterilizer components, adjustment of cycle parameters, a change of packaging material or supplier, a change in product construction, transfer of the manufacturing site, a change in loading arrangement, or a rising trend in bioburden all call for reassessment. Even with no changes at all, requalification should be carried out at the specified interval. The interval requirement is per the current valid version of the standard text.

It is worth building a "change type versus does it trigger sterilization requalification" table directly into the quality system documentation, so the criteria are decided in advance. Otherwise every change turns into an ad hoc meeting, which is both slow and easy to get wrong.

A related cross-check on method selection: if the product tolerates irradiation, the ISO 11137 route often removes the entire aeration and residuals workload; if it tolerates moist heat, ISO 17665 is worth considering. The value of EO lies in its low process temperature and good penetration, which suits heat-sensitive and irradiation-sensitive polymers and products containing electronics.

What to prepare before placing an order

The depth of third-party laboratory involvement varies widely. Some companies contract out only bioburden, biological indicator enumeration and residual testing and write the validation protocol themselves; others want the laboratory to run from protocol design through to the report. Either way, the earlier the following material arrives, the more time is saved.

A few preparation details that are easy to overlook. Samples must be made with the same process and packaging as volume production: validating on hand-sealed samples while production uses an automatic sealer badly weakens the report. Sample quantities should be prepared for the worst case - beyond the units for the formal runs, hold back retest samples, retained samples and the destructive samples for residual testing, because topping up later usually means rescheduling production and re-running the chamber. When you ship, mark the least favourable locations on the drawing so the laboratory does not have to reverse-engineer the construction. And if packaging validation runs on the same batch, put accelerated ageing and distribution simulation on the same Gantt chart as the sterilization validation, since they share samples and jointly determine the submission date. For detailed sample state requirements and shipping instructions, see testing requirements.

About SUNGO Lab

SUNGO Lab is a third-party medical device testing organisation accredited by CNAS, CMA and IAS (USA), with laboratories in Shanghai and Hefei. To be clear, any accreditation mark only demonstrates the technical competence of the laboratory within its accredited scope; it is not a commitment as to market access in the target market. On ethylene oxide sterilization we can take on bioburden determination, biological indicator related testing, sterility testing, ethylene oxide and ethylene chlorohydrin residual testing, as well as packaging integrity and seal strength, biocompatibility evaluation and other supporting items, and we can help work through the structure of the validation protocol and how the submission documentation should read. The specific differences between Chinese registration, CE and FDA filings are also worth a conversation before the project starts, to avoid retracing steps.

To scope timelines and cost, call +86 132 4819 8029 or request a quote. Send us the product construction drawing, the packaging format and the target markets, and we can come back with a test item list and a proposed schedule.