Bioequivalence and Bioavailability Forum

Dear Charl,

we must not consider validation as an end in itself, but as the

“demonstration of the method’s suitability for use”.

We expect minimum concentrations (i.e., which will allow a reliable coverage of 80% of AUC_∞) → we develop a method suitable for achieving this goal → we validate our method to demonstrate this assumption.

If during the course of a study you find out that your method is insufficient, IMHO you have two options:

1. stop analysis, revalidate (e.g., for a lower LLOQ), and re-analysis
   
2. simply ignore it; subjects with insufficient data may have to be excluded by the statistician from assessments

If you opt for method #2, you are risking insufficient power (e.g., in BE), or a drop below the minimum sample size (e.g., in BA- or PK-studies).
Please note: IMHO, it’s not a good idea to go with method #2, discover bioINequivalence, and afterwards switch back to method #1. Regulators may interpret such a procedure as data mining. 
Therefore, I would recommend a blinded plausibility review as soon as possible (i.e, already during bioanalytics; not after its completion).

See also this post and followings.
The topic of values between LOD and LLOQ is leading to controversial discussions approximately every second month at David Bourne’s PKPD-list.
To get an overview use the search function; suggested keywords: ‘LOQ’, ‘limit of quantification’, etc.
There was a rather lengthy discussion in May/June 2007.