Bioequivalence and Bioavailability Forum

Dear all,

❝ Maybe I’m wrong, but I got the impression from Harish’s post that you are talking about varying responses of single measurements within an analytical batch.

❝ Ohlbe, I’m not sure whether you are talking about duplicates?

I was talking about single measurements, where you sometimes have an IS response which is very different from what you have for other samples (for instance half the "usual" response).

❝ If measuring singlets, we don't know whether or not the ratio is affected; we only may notice an 'unusual' IS response.

True, and that's really a complex issue.

Actually I've come across different situations, which may have different causes.

One situation is when all samples from one or several given subject(s) have an IS response which is significantly lower, or higher, than the response of the calibration and QC samples. I've seen runs where subject samples all had an IS response 20 to 30 % lower than cal and QCs. There you may suspect matrix effect (this was LC/MS/MS) and investigate whether it affects your analyte and your IS in a similar way, and thus whether the ratio is modified or not.

Another situation is isolated IS "abnormal" response. HS, you have given a number of reasons which may result in a decrease in IS and may, or may not, affect the ratio. This makes it important to note any abnormal event during sample processing (such as a blocked SPE column). The only way to check whether the ratio was affected or not is to repeat the assay for that sample... The question is: how much IS variation can we accept and what would be the limit to trigger a reassay ? As already discussed, there are different ways to deal with this, for instance:

• define criteria in a general SOP (for instance, reassay the sample if the IS peak area differs by more than (50 % ?) from (the mean IS peak area in your calibration samples ? in your calibration and QCs ? In the run ?) (many possibilities !)
  
• define criteria in your method SOP, based on the variations during method validation, and the impact they had - or not- on the accuracy of the affected samples.

Whatever you do, should be done consistently: I've seen runs where samples were reassayed for "IS variation", but in which samples with a higher variation were not repeated. Or failing QCs not reported due to "IS variation", while subject samples which had the same variation (or worse) were accepted and reported...

Regards
Ohlbe