Scenario - Sterility Test Failure

Your QC Micro team says they have a sterility test showing growth in the test plates.

First thought

• Patient risk - A failure with product sterility leads to an adulterated product and potential patient harm - a failure here might indicate a failure with batches already certified

• Normal batch for release or stability sample?

  • Batch for release - everybody be cool, place batch on quarantine. Possibly quarantine other batches made in same area. Document an initial risk assessment to keep the line filling or not.

  • Stability test - risk profile is larger - this batch is on the market and is potentially defending quite a lot of product also on the market. Document an initial risk assessment to regarding product on the market.

• Their is a strong possibility this is a false positive. Mainly because you have to work hard to truly fail a sterility test. (More below)

• The initial risk assessments could be short - a paragraph - but you need to understand your core rational for acting or not acting. This can be updated as you go.

How would you manage this?

Route 1 - Obvious error in the lab test

For sterile products, the sterility test is a mandatory product release test and as with the case with any test (or reoccurring process) there will be failures.

The core QMS process would be a OOS record (or a deviation depending on your QMS structure) to document the initial results, analysis, look for root cause and plan corrective actions. If you work in a company making sterile product you will have a microbiological OOS procedure in place.

This will include all of the obvious steps. You will obviously expect your microbiology team to be executing this but its useful to know what is typically checked.

• Identification to the species level of the recovered isolate

• Check for obvious errors during testing

• EM conditions of the test chamber

• Personnel dabs

• Training level and experience of test microbiologist

• Results from test performed on the same day

• What isolates may have been present for testing in the area before the test was carried out

• Have routine cleaning actions been performed?

• Have sterility checks for all relevant reagents passed?

It may be possible to identify an error during testing but be wary, there are only four conditions under which the initial failing result may be invalidated, and you can only repeat the test if the results are judged invalid. The European Pharmacopoeia restricts criteria to one or more of the following conditions only:

“(a) the data of the microbiological monitoring of the sterility testing facility show a fault

(b) a review of the testing procedure used during the test in question reveals a fault

(c) microbial growth is found in the negative controls

(d) after determination of the identity of the micro-organisms isolated from the test the growth of this species or these species may be ascribed unequivocally to faults with respect to the material and/or the technique used in conducting the sterility test procedure”

So simply put you have to:

• Find the exact same bug in the area (EM plates or dabs) or,

• Prove an obvious error such as a cracked container/broken glove, or

• Find the same bug in the negative controls - suggesting an issue with either technique or a common reagent.

In other words to invalidate you need good testing records with a micro team that pick up an obvious error and capable of positive isolate identification - morphology or genus level identification simply put wont be enough.

In these circumstances you will be invalidating the test and performing a repeat (if the batch size is over 500 units, that’s another 20 containers, if one more fails - no further retest).

Route 2 - Genuine Failure.

If you are incapable of invalidating the test result and have no root cause in testing you need to treat the results as valid.

From a QMS perspective you may transition from a microbiology OOS at this point to a Deviation record. This is a whole team effort and will involve several departments. This is likely to have four main features:

1. A thorough root cause investigation. You and the team will need to review a lot of data including:

• The contamination control strategy for the production area including

  • Sterility controls such as autoclave reports, depyrogenation records, filter integrity test results

  • EM Data including routine monitoring and periodic testing such as HEPA integrity tests

  • Sterility process biannual validation reports

  • Preventative and Unplanned Maintenance records for the area

  • Batch records, filling intervention logs

  • Training records and the date of the last APS

  • Some particular failure modes become more of a concern for a stability batch - has the sealing (Container Closure Integrity) degraded with time? You may want to challenge the seal integrity of additional stability samples.

  • Sterility controls for bought in sterile materials (think irradiated AS / excipients / containers)

2. Wider communication

• At this point you may need to notify a marketing authority or a customer if the failure relates to a sterility test for a stability sample.

• Senior management will likely appreciate a heads up…..

3. Risk assessment revisit

• The root cause investigation will inform the extent of any risk that emerges and the risk assessment will need to be constantly reviewed.

• The risk assessment will likely need summarised for the communications outlined above.

4. Batch disposition

• OOS, valid result, batch rejected. Possible recall if a stability sample.

Route 3 - Valid, non genuine failure.

A third possibility does exist in this scenario. You may, through diligent investigation and a robust risk assessment, conclude that the test failure is not a genuine sterility failure. Some experimental approaches may support this conclusion.

• If the product in question has either physical properties, or preservatives likely to inhibit growth or kill the recovered isolate - then it is unlikely that the isolate made it alive from your manufacturing environment into your test. However this must be viewed with two further caveats

  • You will need to demonstrate this with samples from this batch of product (ruling out any batch specific issue with these inhibitory features) and this isolate strain (proving to a degree that you haven’t identified a new strain of product resistant microbe) by incubating a large amount of the later in the former. This shows your product would kill the isolate before it got to the test plate.

  • This experiment is only for additional information and it will only provide risk assurance for other batches of product - as per the note on the pharmacopeia above this is not an acceptable way to invalidate a failing test.

• While you are not permitted to do a retest if you cannot invalidate the results, the batch is going to the bin, you can test additional samples for information gathering purposes. Again this has limitations in the information it provides

• The sterility test itself, and any additional sample of a practical size - is only going to detect gross contamination in a batch. It dos not provide assurance of product sterility.

In this case you are documenting the full investigation as per the second option and a risk assessment that gives you confidence it is not a genuine fail. The batch of product is still dead most of the time (see below).

Route 4 - Special (and unlikely) cases

There may be a rare occasion where you confirm a sterility fail and the batch is still released. This will only happen in the following circumstances that I am aware of.

• The product is necessary to save the life of a patient and no alternative exists within the required timeframe - the only example that comes to mind here is an ATMP that makes use of the patients own cells (Autologous treatment). Possibly a vaccine in time of pandemic for something as bad or worse than COVID-19 (speculation).

  and

• The marketing authority gives you permission to release an OOS product - something a QP cannot do as per Annex 16. This is a batch specific variation or dispensation.