Bioequivalence and Bioavailability Forum

Hi A.Shtaiwi,

❝ […] would it be acceptable to give all subjects the fasting treatments first (Periods 1–2) and then the fed treatments (Periods 3–4)?

Acceptable, yes. However, IMHO, not optimal (see below).

❝ Or do we need to randomize subjects so that fasting and fed are mixed across periods?

No.

❝ A = Test, fasting

❝ B = Reference, fasting

❝ C = Test, fed

❝ D = Reference, fed

I suggest to reverse the order of your option 3 and go with

S1: C → D → A → B
S2: D → C → B → A

Why? In many cases it is more difficult to demonstrate BE in fed state than in fasting state. If you have prior information about the respective variabilities, you could include more subjects for periods 1–2. You could also opt for a sequential design in each of the parts. For details see this presentation.

To consider:

• Don’t perform an ANOVA of pooled data. Not only the variances of the parts will be different but very likely the estimates will be biased as well. Evaluate the fed and fasting parts separately. This is also recommended in ICH M13A:
  • In studies with more than two treatment arms, e.g., a four-period study examining fasting and fed conditions […], the analysis for each comparison should be conducted excluding the data from the treatment arms that are not relevant for the comparison in question.
    
• The assessment of food effects of test (C/A) and reference (D/B) is only ‘nice to know’. Exclude the respective other conditions to get incomplete block designs and evaluate them as paired groups (assuming equal period effects).

❝ Which approach would regulators (e.g. EMA) accept?

Likely any. As of today ICH M13A is implemented by Swissmedic (Switzerland), the FDA (US), the EMA (EEA), the MHRA (UK), the JFDA (Jordan), and will be implemented by Health Canada in December 2025. See there for the current implementation status.