Bioequivalence and Bioavailability Forum

Hi Darborn,

❝ So the point is under the circumstance that I mentioned, the bioequivalence is still valid …

As long as the 90% confidence interval lies entirely within the acceptance range, bioequivalence is accepted with \(\small{\alpha \leq0.05}\). In other words, the patient’s risk of inequivalence (i.e., the Type I Error of erroneously accepting BE) does not exceed 5%. That’s the only issue regulators should be concerned about. Whether or not the CI includes 100% is irrelevant. For some historical background see this article.
Furthermore, there is no ‘robustness of BE’ (as you wrote in the subject line of your original post). The outcome of a comparative BA study is dichotomous (pass | fail): Either BE was demonstrated or not.

❝ … but the clinical relevance may need justification.

While the choice of the clinically relevant difference \(\small{(\Delta)}\) is essentially arbitrary, for ‘normal’ drugs (i.e., not NTID or HVD) \(\small{\Delta=20\%}\) was proposed already in the 1970s^1–3 and is the globally accepted standard for more than thirty years. Based on that, we get the ac­cep­tance limits \(\small{\left\{\theta_1,\theta_2\right\}}\) by \(\small{\left\{100\times\left(1-\Delta\right),100\times1/\left(1-\Delta\right)\right\}=
}\) \(\small{\left\{80\%,125\%\right\}}\), which have to be pre­defined in the protocol. Although \(\small{\Delta}\) is not stated as such in a guideline (e.g., the most recent one⁴), it can be back-calculated from the acceptance limits as \(\small{\Delta\%=}\) \(\small{100\times\left(1-\theta_1/100\right)=}\) \(\small{100\times\left(1-100/\theta_2\right)}\). Therefore, a justification is not needed.

Although bioequivalence is not a science in the Popperian sense, we have decades of empirical evidence that the concept works. If an agency doubts it and demands a ‘justification’ specifically for your study, it is their (‼) damn duty to prove that the globally accepted approach is flawed.

»Extraordinary claims require extraordinary evidence«⁵

❝ Am I right?

First part yes, second no.

---

1. APhA. Guidelines for Biopharmaceutic Studies in Man. Washington D.C. February 1972.
   
2. Metzler CM. Bioavailability—A Problem in Equivalence. Biometrics. 1974; 30(2): 309–17. doi:10.2307/2529651.
   
3. Westlake WJ. Design and Evaluation of Bioequivalence Studies in Man. In: Blanchard J, Sawchuk RJ, Brodie BB, editors. Prin­cip­les and perspectives in Drug Bio­avail­abi­li­ty. Basel: Karger; 1979. ISBN 3-8055-2440-4. p. 192–210.
   
4. ICH. Bioequivalence for Immediate Release Solid Oral Dosage Forms. M13A. Final version. Geneva. 23 July 2024. Online.
   • 1.2.1 Bioequivalence
     Two drug products […] are considered bioequivalent if their relative bioavailability (BA) […] lies within acceptable predefined limits.
     2.2.3.1 General Considerations
     The assessment […] is based on 90% confidence intervals for the geometric mean ratios (test/comparator) […]. This method is equivalent to two one-sided t-tests with the null hypotheses of bio­in­equi­valence at the 5% significance level. The PK data should be transformed prior to analysis using a logarithmic transformation.
     2.2.4 Bioequivalence Criteria
     The 90% confidence interval […] should lie within a range of 80.00 – 125.00%.
     (my emphases)
5. Sagan C. Broca’s Brain: Reflections on the Romance of Science. New York: Random House; 1979. ISBN 0-345-33689-5. p. 73.