Decontamination of isolator materials

Aseptic filling of parenteral drugs needs the highest microbiological quality of container, product and production environment. Positive pressure barrier isolators separate the filling process form operator environment, which is a known source of potential microbiological contamination. Due to the effectiveness of separation, isolated filling lines can be placed in a lower grade air quality environment (at least grade D according to European GMP).

After preparation of the filling equipment for production with doors open, the inner surfaces have to be bio-decontaminated. Therefore hydrogen peroxide (H2O2) vapour is a well accepted biocidal agent, which has to be generated from liquid hydrogen peroxide solution and distributed homogeneously in the aseptic area. The treatment process parameters for this need to be developed and validated.

As an indicator for the efficiency of the process, coupons made from stainlees steel which are then inoculated with ? 1.000.000 spores of Geobacillus stearothermophilus are produced.

The largest surface area inside an isolated system usually comprises stainless steel. But there are many more materials, such as glass, elastomer materials for gloves and seals, plastic materials for size parts, which have surface properties that differ from stainless steel. The purpose of this study was to evaluate and quantify the influence of these materials on the inactivation of bioburden.

For the quantification of the inactivation effect D-value determination was chosen. This refers to decimal reduction time, ie, the time that it takes at a certain temperature to kill 90% of the organisms being studied. Thus after an organism is reduced by 1 D, only 10% of the original organisms remain.

A choice of about 50 materials that are used for construction of isolated fill lines, or that could be used or that are of specific scientific interest were selected. The surface quality/machining finish had to be identical to the relevant surface for an isolated line.

Samples of these materials were cut into pieces and sent to Apex Labs, Sanford, NC, USA, for cleaning and inoculation. The inoculation was performed with a spore suspension that was identical to the suspension used to manufacture the commercial stainless steel BIs, which were carried along through all tests. Each BI was placed in a Tyvek envelope to ease handling and promote comparability with BIs used for cycle validation.

The D-value determination was performed in a production-size isolator of 7.5 m3, manufactured by Bosch. For generating hydrogen peroxide gas, 35% medical grade solution sourced from Merck KgaA, Darmstadt, Germany was evaporated with a Bosch SafeVAP bio-decontamination system.

The bio-decontamination cycle consists of 4 steps:

• Leak test and dehumidification of isolator inside
• Injection of H2O2-vapour with high injection rate (22 g/min) in order to achieve a high peroxide concentration (e.g. 800 ppm)
• Keeping the peroxide level over a certain time in order to complete inactivation (reduced injection rate: 13 g/min)
• Aeration with high air volume to remove all peroxide from the isolation system.

The read out (growth / no growth) took place every day. The result after seven days was analysed following the Holocomb Spearman Carber Procedure to calculate the D-value of the specific material surface.

In a pre-study the selected materials were studied with reduced statistics (3 per take-out, interval 5 min) in order to get an estimate of the fractional window.

In the main study, the fractional window was tested with more statistics (5 per take out interval every 3 – 5 minutes) in order to determine the D-value with greater accuracy.

In an additional study materials of comparable D-values (coming from the main study) were exposed together in order to cross-reference between these materials and exclude deviations coming from specific cycle behaviour.

The results of these investigations offer some surprises as well as the confirmation of results known from literature. Only a few materials exhibited a very negative impact on kill (silicone rubber VMQ, PVC-U, fluorocarbon rubber FPM). Some materials had D-values that were much lower than reported in the literature (acetal polyoxymethylene copolymer or POM, aluminium). And not fully understood at this point of investigation was the high D-value obtained from electropolished stainless steel – compared with other stainless steel surfaces, which had much lower D-values.

A second set of materials was investigated with two aluminium qualities and surfaces, which were anodized with different thickness of the oxidative layer (same Rz – roughness however). The thicker the anodization layer, the worse the kill rate. And cast aluminium exhibited lower kill ability than forged alloy aluminium. While some aluminium surfaces showed kill D-values that would be reasonable for an isolator application, aluminium has major disadvantages in respect of hygienic design and therefore is not feasible for use within isolated filling lines.

An important insight from this study was that not all the materials used or potentially used in isolators will guarantee an effective inactivation of bioburden and those materials should be excluded, therefore, from the construction of filling machines and isolators.

Due to its reproducible properties, commercial availability and taking into account the fact that stainless steel represents the largest surface inside an isolator, coupons made from this material should be the standard indicator for performing cycle development and validation. But for defining the safety margin (extra time in the bio-decontamination phase (III) to have complete kill under all circumstances) it is reasonable to carry the material with the highest D-value through - at least - the final steps of the bio-decontamination cycle development.

Artefact study

In a second series of investigations, different techniques were applied to learn more about the effect on certain artefacts.

Material surfaces with an extremely high D-value or incomplete kill within the tested take-out intervals were suspected of formating multiple layers from 6 log inoculation with spores. Therefore, additional samples with 4 log inoculation were manufactured and tested in the same cycle. In general 6 log BIs exhibit higher D-values than 4 log BIs of the same material. In particular, 6 log inoculated PVC-U samples exhibited a much higher D-value (a difference of 3 min.) compared with the 4 log samples of the same material. This is a strong indicator for the multilayer hypothesis.

The D-value difference of most of the other materials investigated with 4 log and 6 log inoculation was within the spread of microbiological precision, so no artefact effect could be stated.

Stainless steel commercial BIs were exposed - with and without a Tyvek-wrap - to the same cycle, in order to determine the additional challenge provided by applying wrap a to the BIs. The D-value of wrapped BIs was exactly double the D-value of unwrapped samples. Due to handling (re-contamination risk) it is recommended to use wrapped BIs for cycle development. The additional challenge is part of the safety margin, which helps to reduce any risk of microbiological contamination.

Surface topography

All materials were examined by scanning electron microscopy (SEM) to study the influence of specific surface topography on inactivation behaviour. A qualitative correlation between topography and D-value could be found for many materials. Electropolished steel samples provided a surprise: SEM study revealed scratches that could hide spores from exposure to a sporicidal agent. The scratches were not present immediately after electropolishing, but very likely happened during the cutting of the samples for sample preparation. This could explain the high D-value result of the electropolished stainless steel.

Roughness determination

To understand the effect of topographical surface on D-value roughness, white light interferometry measurements were applied to all samples. As a result, a correlation of surface roughness with D-value could be seen with many of the materials, but not all. Therefore additional effects need to be taken into account.

Contact angle

The impact of inoculation on kill behaviour was studied by determination of the contact angles of sessile drops, made from the spore suspension solvent used for inoculation, on top of the material surfaces. There is a clear correlation between wettability and D-value for most of the samples (the higher the contact angle, the smaller the inoculation drop, the more likely the formation of multi-layers – the higher the D-value. But again, there are more influences than contact angle effects only.

Recovery rates

According to international standards recovery rates have to meet a certain spread for validatability of the results. This study resulted in a recovery ranging from 58.4 to 89.7 %, which is well within acceptable spread (50 – 300%).

To exclude inactivation effects on growth of potential survivors by carrying H2O2 with the material into the nutrient media, blank materials – ie, with no inoculation – were exposed to a bio-decontamination cycle of maximum length and transferred applying the same techniques into nutrient media, which was inoculated with 2 log spores after transfer. In parallel, the same number of blank material samples – not exposeed to the cycle – were handled following the same procedure.

EPDM (ethylene propylene diene monomer) showed an inactivating effect on growth without seeing any peroxide, that means that the material itself has an inhibiting property. For nitrile rubber (NBR), poly amide (PA) hose and poly urethane (PU) belt an inhibition was observed only for the decontaminated samples. All other materials showed no retardation of the outgrowth of the test organisms.

D-value reproducibility

Through all the cycles, commercial available standard stainless steel BIs were tested as a reference. Therefore D-value determination was performed about 50 times with the same batch. The D-values were spread between 0.2 min (min) and 0.9 min (max), mostly oscillating around 0.5 min, which was the average D-value calculated from all cycles.

Curious to learn if there was an effect of the H2O2-concentration on the D-value, the researchers compare these results with H2O2-concentration measured by a NIR probe. But no correlation could be found between measured ppm and D-value. It should be noted that the spread between maximum and minimum concentration measured in that set of experiments is within the accuracy of the NIR-measurement.

The results of this material study show different inactivation properties for bioburden for different material surfaces. As a result some materials should be avoided in the construction of isolated filling lines.

It is a safe and reasonable approach to use 6 log inoculated stainless steel coupons, wrapped in Tyvek as standard BIs for performing cycle development and cycle validation. Additional tests with the material of construction showing the highest D-value can be recommended for determining the safety margin.

Although there is an influence from the entire surface quality, the biodecontamination system together with the process philosophy applied, many of these findings are of general interest and could be used to cross-referencefuture studies with a different experimental setup.