Bioequivalence and Bioavailability Forum

Dear Pavan!

❝ suppose the percentage ratio of AUC_0-t and AUC_0-inf is less than 80%.

80% of ‘covered’ AUC are literally given in only some guidelines (e.g., nothing is mentioned in FDA’s at all). Mainly they speak about ‘sufficient long sampling to get a reliable estimate of the elimination phase’,

A current document comes from EMEA, stating:

Provide information regarding size of extrapolated area. The extrapolated AUC should not be higher than 20% in any subject (not only the mean extrapolation).

An older one from ANVISA (2003):

The AUC_0-t must be equal or greater than 80% of the ASC_0-inf, except in the cases where the truncated AUC is used;

This requirement is still given in 2006’s document (in Portugese).

EMEA’s NfG (2001) states:

The sampling schedule should be planned to provide an adequate estimation of C_max and to cover the plasma concentration time curve long enough to provide a reliable estimate of the extent of absorption. This is generally achieved if the AUC derived from measurements is at least 80% of the AUC extrapolated to infinity. If a reliable estimate of terminal half-life is necessary, it should be obtained by collecting at least three to four samples during the terminal log linear phase.

EMEA’s BE Draft (2008) states:

For single dose studies, the percentage of AUC_inf that is covered by AUC_t should be reported for each subject in each period if the observation period is shorter than 72 hours. Subjects should not be excluded from the analysis on the basis of this calculation, but if the percentage is less than 80% in more than 20% of the observations then the validity of the study could be questioned.

❝ What is the justification we can give?

Stop! Think! Act! Don’t perform the BE-assessment according to the protocol! And document clearly that you haven’t done it yet. Otherwise regulators might get the impression any strategy you are developing to deal with the problem was data-driven (i.e., you played around with different methods and selected one which favours BE).

Some thoughts:

• If high residual areas are seen in the majority of cases and AUC_0-∞ is a primary metric in the regulation (FDA), you are in trouble.
  
• Since half lives were longer than anticipated in study planning, check for predose concentrations in period II.
  
• You may go with a truncated AUC-approach (not extrapolating at all). Search the forum on the topic.
  
• If you see high residual areas just in a few subjects (both with test and reference), you may have discovered a subpopulation of slow metabolizers. Check the literature.^* Exclude them and perform the analysis of the reduced data set as primary and the full data set only as a supportive sensitivity analysis. Discuss the differences.

---

• Things like that may happen even for well-documented drugs (more than 50 years of marketing experience). Patrick et al. reported a novel methylphenidate-poor metabolizer of a previously undocumented phenotype. Concentrations in this subject were found to be 100times [sic] those of the other 19 subjects in the study:
  Patrick KS, Straughn AB, Minhinnett RR, Yeatts SD, Herrin AE, DeVane CL, Malcolm R, Janis GC, Markowitz JS. Influence of Ethanol and Gender on Methylphenidate Pharmacokinetics and Pharmacodynamics. Clin Pharm Ther. 2007;81(3):346–53.