Bioequivalence and Bioavailability Forum

Dear all!

I agree – almost.
See also the EMA’s GL (4.1.5 Characteristics to be investigated, Pharmacokinetic parameters):

Actual time of sampling should be used in the estimation of the pharmacokinetic parameters.

At the recent EGA-symposium on BE we had a little discussion about AUC₇₂. Imagine a situation, where in some subjects the value at 72 h is <LLOQ, and the previous sampling time point was 60 h. If the ‘true’ relative bioavailability is 100%, in balanced studies we face an equal chance for both formulations to show values at 72 h below the LLOQ. Overall we would get an unbiased estimate comparing AUC₆₀ to AUC₇₂. But if the ‘true’ relative bioavailability is ≠100%, the PE will be biased away from 100%.
Example: the true value is 95% – we will expect more subjects after the test to show values below the LLOQ than after reference. If we base our decision on individual AUC_t-ratios, the estimate will be <95%. In other words, the patient’s risk will not be affected (that’s the reason why regulators are not worried about apples and oranges), but the producer’s risk will increase.

See also these threads:

• #1334
  
• #2323

Conclusion: Keep deviations as small as possible and try to set up the bioanalytical method in such a way that you are able to measure the last concentration.