Bioequivalence and Bioavailability Forum

Dear sciguy!

❝ The sponsor wants to compare two different formulations of a drug with 2 different strengths, say 5 mg vs 10 mg. Dose normalization on the relevant PK parameters is performed and the assessment of equivalence using standard BE criteria is applied. If BE is met, we can only conclude dose proportionality but not bioequivalence since 2 different strengths are being compared. True bioequivalence must be evaluated on non dose-normalized data. Agree/disagree?

An often overlooked trap. Let’s consider the basic BA equations:

        DT × FT              DR × FR
AUCT = ————————— and AUCR = —————————
          CLT                  CLR

If we assume (!) DT ≈ DT and CLT ≈ CLR these terms cancel out and we obtain

      FT     AUCT
BA = ———— ≈ ——————
      FR     AUCR

If we dose-adjust and want to assess for BE we try to kill two birds with one stone. In your example comparing twice the AUC after the 5 mg dose to the AUC after the 10 mg requires strict linear PK: Both Clearance and fraction absorption independent from dose (a straight line through the origin). Whereas the former is drug-specific the latter might be (also) formulation specific. Imagine a hypothetical scenario (of course nonlinear PK): capacity limited absorption and Clearance. At 5 mg F = 0.9, CL = 12.5 and at 10 mg F = 0.8, CL = 10. AUCs are 0.360 and 0.800. Dose normalisation will lead to BA = (0.360 × 2) / 0.800 = 90.0 %. So far so good – but is this ‘true’? What we want to compare in BE is F_T/F_R. Only because Fs are not accessible we use AUC_T/AUC_R instead. F_T/F_R from above would be 112.5 %. Even if I have strong evidence of linear PK (where dose-normalisation would definitely work) I would opt to dose the same amount if BE is to be assessed.

P.S.: There’s another thread dealing with dose-proportionality. I will elaborate on it later on over there. Just a first glimpse: We once had a study where an MR formulation (20 mg oad) was compared to an IR formulation (2 ×10 mg, τ 4 h). AUC was BE. Since no data on dose-proportionality were available we dosed the MR formulation also at 10 mg and 40 mg (4-way crossover). We showed BE after dose-normalisation (strict linear PK) after applying a Bonferroni-correction (α_adj 0.0167 ≡ 96.67 % confidence intervals). It was a hybrid application and we had two clinical studies showing safety/efficacy at these levels (compared to reference and superiority to placebo). European regulators accepted the study (= approval for all three strengths) but told us that they didn’t like our approach. They would have preferred to get three studies instead: 2×10 mg MR vs. (10+10) mg IR, 1×20 mg MR vs. (10+10) mg IR, 1×40 mg MR vs. (2×10+2×10) mg IR… (only the 10 mg IR acting as reference).