Bioequivalence and Bioavailability Forum

Dear Ravi!

❝ Suppose we have three formulation of a drug product (T1, T2, T3) in different dose and one reference product (R). We can check for BE between Test and reference product.

Let’s assume your reference product is dosed with 100 mg, T₁ with 25 mg, T₂ with 50 mg, and T₃ with 100 mg. If dose linearity (not proportionality!) is known from the literature or previous studies you may show BE of 4×T₁=R, 2×T₂=R, and T₃=R.
But:

• You may run into multiplicity issues; with three simultaneous comparisons at a level of 0.05, since the uncorrected error type I (patient’s risk) may increase to 1-(1-0.05)³=14.26% instead of 5%. Therefore it may be necessary to go with an adjusted α level of 0.05/3=0.167 or an confidence interval of 96.67% instead of 90%. The patient’s risk is kept <5% with 1-(1-0.05/3)³=4.92%. You will have to increase your sample size accordingly to maintain the desired power (see the example below).
  
• The definition of dose linearity is a battleground. Most people use a power model (PK response = A · dose^B), but there is not agreement about the goalposts of ‘allowed’ deviations of the exponent B from 1, not speaking about confidence intervals, necessary power, etc. You can avoid that by dosing test formulations at the same level as the reference. This approach is preferred by some regulators.

❝ Now my question is can we check for Bioequivalence between T1, T2 and T3 knowing that all three test products are of different dose strength.

Why? Are you trying to demonstrate dose linearity in the same study? You will increase the number of simultaneous comparisons even further from three to six (T₁=R, T₂=R, T₃=R, 2×T₁=T₂, 4×T₂=T₃, 2×T₂=T₃). α level adjustments will call for testing at 0.05/6=0.00833 increasing the sample size.
Example: Expected deviation test from reference -5%, CV_intra 20%, power 80%; 1 comparison (conventional 2×2 BE study) n=20, 3 comparisons (α 0.05/3) n=28, 6 comparisons (α 0.05/6) n=32.