Bioequivalence and Bioavailability Forum

Dear Nees!

❝ I would like to know how to calculate AUCreftmax?

1. Calculate the median of t_max values observed in all subjects under the reference treatment.
   
2. Calculate partial AUCs from t=0 to this time point for all subjects under all treatments.
   This is tricky, because it’s unlikely that you have a sampling time point at this value in all cases. Therefore you have to interpolate between two adjacent time points in these cases. A reasonable approach is to use linear interpolation if the truncation time point ≤t_max and log/linear otherwise. State the procedure in the protocol.

❝ Why we want to calculate this parameter?

Because in the US (called “early exposure”) and in Canada this metric is mandatory if considered clinically important (rapid onset of action or adverse effects related to absorption rate). In the US the 90% confidence interval has to be within 80%-125%, whereas Canada requires only the T/R-ratio within 80–125%. FDA’s requirement is very difficult to meet because the early parts of the AUC are quite often highly variable (even for ‘nice’ formulations)…

In Europe this approach was discussed (see this post and followings), but is not required in the current guideline.

Hi Neethu,
If you have a WinNonlin then requesting the partial area to a common Median Tmax as a partial area will interpolate with your desired choice of linear/log trapezoids.

Incidentally it's relatively easy to make a workflow in Phoenix to select an AUC to an Individual's Tmax using the Join function if you have to do that too,



Simon.