Bioequivalence and Bioavailability Forum

Dear all,

some comments on this week’s workshop in Bonn.

»The BfArM still sleeping at 5 a.m.«

---

1. SD vs. MD

Status:

• The U.S. FDA generally does not require MD studies (exception: studies in patients under therapy due to unacceptable safety profile in healthy subjects).
  
• Japan’s NIHS requires SD studies, although “multiple dose studies may be employed for drugs which are repeatedly administered to patients.”
  
• Brazil’s ANVISA does not require MD studies.
  
• Canada’s HPFB/TGD requires MD studies only if substantial accumulation is expected, i.e.,
  AUC_t > 80% AUC_∞.
  
• The EMA currently requires MD additionally to SD in all cases.
  
• According to the draft MD studies might be waived under certain circumstances:
  • AUC_t > 90% AUC_∞ and
    
  • BE demonstrated in SD for additional PK metrics, e.g., pAUCs (C_τ under discussion).

Obviously the EMA does not trust in the empiric evidence of no problems in other countries (“absence of evidence is not evidence of absence”). The leap from ‘always’ to ‘never’ is not for cowards. Like in many other places the drafting party followed Canada’s rules, but not without tightening the thumbscrews.
As Alfredo García-Arieta pointed out “The 10% residual AUC are arbitrary; we could have used 20% or 5% as well.” OK, Canada is successfully applying the 20% for more than two decades [sic] now…
I love arbitrary values.

The suggested procedure starting with the SD study is as following:

• If AUC_t ≤ 90% AUC_∞ target metrics are AUC_∞, AUC_t, and C_max. Whether AUC_∞ is primary for controlled release and AUC_t for delayed release is under discussion. pAUCs for multiphasic profiles.
  MD mandatory (waiving criterion not fulfilled).
  
• If AUC_t > 90% AUC_∞ target metrics as above + additional metrics describing the shape of profiles (pAUCs, t_75%?).
  • If additional metric(s) pass BE, a MD study might be waived.
    
  • If the additional metric(s) fail, a MD study is required.
• If the MD study passes with the conventional metrics (AUC_τ, C_max, C_min), the fact that ‘shape’ metric(s) failed in SD will not lead to rejection of the application (rationale: BE demonstrated with conventional metrics both in SD and MD).

Now it is up to us to challenge the 90% criterion (empiric evidence from Canada). Furthermore, BE of C_τ makes sense, IMHO. Will be a tough job to convince Alfredo García-Arieta claiming: “I don’t believe in simulations.”

He was asked how achievement of steady state can be demonstrated and whether a statistical test is required. The answer was “A test is not expected. Reporting of pre-dose concentrations and plots are sufficient.” All other members of the drafting party nodded in agreement.

---

2. _partialAUC (if applicable)

Status:

• The U.S. FDA requires pAUC_Tmax,ref as a measure of early exposure if clinically relevant. T_max,ref is the population (‽) median. Conventional acceptance range.
  
• The U.S. FDA requires pAUCs in some product-specific guidances (methylphenidate: ER capsules, ER tablets; zolpidem ER tablets). The truncation time point ‘T’ is a priori set based on PD and might be different for fasting/fed or applicable to one state only. Example: MPH (3 h fasting, 4 h fed), zolpidem (1.5 h fasting only).
  Required metrics: AUC_0-T, AUC_T-t, AUC_0-∞, and C_max; conventional acceptance range.
  
• HPFB/TGD requires pAUC_Tmax,ref as a measure of early exposure if clinically relevant. T_max,ref is defined as the individual’s value. Relative mean within the conventional acceptance range (no CI).
  
• EMA Q&A Ref.4 (02/2012): Pre-specified truncation time point. AUC_0-T, AUC_T-t, C_max,0-T, C_max,T-t; conventional acceptance range.
  
• MR draft (lines 787–792): Pre-specified truncation time point. AUC_0-t, AUC_0-∞, C_max, t_max, AUC_0-T, AUC_T-t, C_max,0-T, C_max,T-t, t_max,0-T, t_max,T-t. Yippee! Ten characteristics. That’s a lot. IMHO, only descriptive.
  Statistical evaluation (lines 800–805): AUC_0-t, AUC_0-∞, C_max, AUC_0-T, AUC_T-t; conventional acceptance range or scaling (lines 807–810).

There was a short discussion whether for delayed release AUC_0-T & AUC_T-t and for prolonged release AUC_0-T & AUC_T-∞ would be sufficient (dropping the respective other AUC-metrics). No consensus was reached.

---

Presentations of the workshop will be available for some limited time in a couple of weeks at EUFEPS’ website.