Bioequivalence and Bioavailability Forum

Dear Pandu,
When the Reference drug itself is more variable (e.g. "prazoles")and we are getting a high intra subject CV than we could go for this type of design.
The study design would be partial replicate. All the subjects will receive reference product in two periods and test product in one period (i.e. it will will be a three way cross over study).
After completion of the study, the intra-subject variation between the two reference arms will be calculated.
If the intrasubject variability is less than 30% conventional bioequivalence limits will be calculated and the product will be concluded as bioequivalent if the 90% CI is between 80% and 125% for AUCinf, AUClast and Cmax.

USFDA Non Binding Recommendations for conducting Bioequivalence studies of Lansoprazole

Recommended studies: 2 studies (one under fasting condition & one under fed condition)

Three Type of study: Fasting and Fed.
Design: Single-dose, two-way, crossover in-vivo
Strength: 30 mg
Subjects: Normal healthy males and females, general population
Additional Comments: Available data indicate that this product may be highly variable in the bioequivalence parameters AUC and/or Cmax. You may consider conducting bioequivalence studies using a replicate design approach. These replicate design studies may be analyzed using the reference scaled approach. The reference-scaled approach adjusts the bioequivalence limits of highly variable drugs by scaling to the within-subject variability of the reference product in the study, and imposes a limit of 0.8 to 1.25 on the geometric mean ratio. The within-subject variability of the reference product is determined in a 3-way modified replicate-design study in which the reference product is given twice and the test product is given once. For general information on this approach, please refer to Haidar et al., Bioequivalence Approaches for Highly Variable Drugs and Drug Products, Pharm. Res. 25:237-241(2008).

Hope this information will clear your query.

For details, see Helmut's post on "Reference scaled approach for HVD".