Bioequivalence and Bioavailability Forum

Dear HS,

thank you for your answer.

❝ ❝ These results occur in 20% of the subjects, and for each subject happen in more than one period, and always during the period 1, before the starting of the drug administration.

❝

❝ I’m not sure whether I do understand you correctly. In those 20% of subjects do you get pre-dose concentrations in all periods? Please clarify.

Just to clarify: the 20% of all subjects shown pre-dose concentrations above LLOQ in period 1.
The 12% of all subjects shown pre-dose concentrations in all periods (1+2+3).
The 10% of all subjects shown pre-dose concentrations in two periods (periods 1+2 or periods 1+3).

❝ ❝ Following several bioanalytical check, the results seem to be correlated to an eventual contamination of the pre-dose samples during the clinical phase, rather than to a bioanalytical bias.

❝

❝ I know the term ‘contamination’ is sometimes used – but how should that practically occur? Somebody spiking vacutainers at BD’s factory (sabotage)? The drug leaking from the formulation, sneeking downstairs to the room where the centrifuges are, and slipping into sample vials? I would rather say it’s an analytical issue.

It could be possible that sometimes the contamination of the head of the pipettes causes the samples contamination (assuming obviously that the tips are always changed).
Regarding the bioanalysis, the samples have been analyzed 3 times, using both the 1st and 2nd aliquote; the repeated analysis confirmed the results. Carry over is checked for each run and during the analytical study no carryover was evidenced.

❝ At least we are talking about small concentrations. Are you using LC/MS-MS or a ligand-binding assay? Guidelines call for checking of interferences and the matrix-effect in ≥6 sources of matrix. Even if you were successful in validation it might be possible to be hit in a larger study by pure chance. In the future consider checking more than six sources of blank matrix if you plan for a large study.

I confirm that we used LC/MS/MS method.

❝ According to the GL you should report all data and remove subject’s data of all periods where pre-dose concentrations are >5% of C_max. This procedure should have been stated in the protocol. Was it?

Yes, of course! we included this as criteria of exclusion in the protocol.

❝ ❝ What is the position of the regulatory authorities in these cases?

❝

❝ I’m not a regulator. Personal experience close to nil. Messed up a 36 subject 3 treatment study (two tests, one reference) ten years ago where the wash-out was too short (stupidly we improved the LLOQ by a factor of five compared to a pilot study) and found pre-dose concentrations in the second period in 21 subjects and in the third period in 18 subjects (but all of them <5% C_max). We presented:

1. Evaluation as planned (ignoring pre-dose values).
   

   ❝

2. Evaluation based on the first period only (i.e., as a parallel design).
   

   ❝

3. Estimating λ_z including the pre-dose values. Correcting all concentrations in higher period(s) based on the estimated time-course of earlier period(s).

Both test formulations were BE based on all methods. Study was accepted (mainly based on [2] – study was large enough). Nowadays [1] is enough. [2] is statistically flawed and should not be done according to the GL [“A test for carry-over is not considered relevant and no decisions regarding the analysis (e.g. analysis of the first period only) should be made on the basis of such a test.”]. [3] only for very adventurous people.

Since we have 20% of the pre-dose concentration in period 1, we exclude that the problem is correlated with a short wash-out period. Therefore we think that we cannot use in our case an approach like your to present our results...

Thank you for the example on your experience and for your advices.

Best regards,
Paola