Bioequivalence and Bioavailability Forum

Hi ElMaestro,

❝ I guess I am trying to say that yes I agree sometimes the 80.00-125.00 criterion appears too strict.

Maybe. The origin of these limits lies in the dark ages. However, see these quotes from pre­sen­ta­tions given at the “1^st MENA Regulatory Conference on Bioequivalence, Biowaivers, Bioanalysis and Dissolution” (Amman, 23/24 September 2013):

“Bioequivalence based on the comparison of bioavailability under strict criteria has proved to be the gold standard for the approval of generic medicinal products with ca. 40 years of ex­pe­ri­ence without major incidents.”¹
“No prospective study has ever found that an FDA approved generic product does not show the same clinical efficacy and safety as the innovator product, even when special populations (e.g., elderly, women, severely sick patients) are studied.”²

In other words, even if the concept might not be strictly scientific at all, it seems to “work”. There are some hints that the concept might be too strict (or the other way ’round, less stringent criteria would also serve the job):

• Products entered the market approved by methods which nowadays would provoke only a smirk in any assessor’s face. Examples: FDA’s “75/75-rule”, testing for a significant difference, a­rbi­trarily widening the acceptance range, studies in steady state only or based on a metabolite in order to reduce variability… Where are the failures?
  
• The wide range from lousy study conduct up to plain fraud which slipped through the attention of regulators for years. Remember Ranbaxy? THX to GMP (which seems to be more easy to inspect than GCP) which results in acceptable product performance – even if bioinequivalence would be the “true” result.

Serious suggestions were made in the past to move forward.

• At one of the BioInternational Conferences (Munich 1994) different limits for classes of drugs were discussed. But: Who would have the balls to set them?
  Les Benet also suggested that as part of any NDA the innovator should run a large replicate study and release the CV_W to the public domain. That would have put an end on pilot studies in ANDAs estimating the variability (however, still necessary to get an idea about the GMR).
  
• Reference-scaling for HVDPs is dependent on the CV_WR estimated in particular studies. That’s completely different from the “one size fits all” 80/125-concept. Sooner or later we will have generics of HVDPs on the market which were approved according to different rules.
  Furthermore, reference-scaling is not reversible. In any 2-way cross-over we could switch the coding of T and R. Studies showing BE for T/R would also show “BE” for R/T. Since the pro­ducts’ CV_w must not be identical (an assumption in 2,2,2-studies) – and in fully replicated designs we have all means to assess that* – a study which passes T/R in RSABE/ABEL based on CV_WR must not necessarily pass R/T based on CV_WT as well. Of course one can make an argument that such a comparison is futile, but I find it a little bit disturbing. At least it violates the symmetry required by the WHO: “The procedures should lead to a decision scheme which is symmetrical with respect to the two formulations (i.e. leading to the same decision whether the multisource formulation is compared to the comparator product or the com­pa­rator product to the multisource formulation).”
  The two Lászlós suggested that regulators should collect CVs of reference products and pool the variances. Due to the much higher sample size such an estimate would be more accurate (and precise) than anything one estimates in an actual study. Then regulators should publish a product-specific acceptance range. As a nice side-effect it would end the inflation of the patient’s risk we face right now.

---

1. José A Guimarães Morais
   EMA Perspectives on BE Regulations
   
2. Leslie Z Benet
   Why Do Bioequivalence Studies in Healthy Volunteers?

---

• For incomprehensible reasons EMA in their holy Q&A-document prefers to ignore the in­for­ma­tion which easily can be obtained in replicate studies. At the days of population-/individual-BE these designs were developed exactly for this reason: To assess different variabilities of test and reference (and the magic subject-by-formulation interaction). In following EMA’s “Methods A & B” we are forced to assume a common variance of T and R (like in 2,2,2-studies) and “Method C” – which would take the entire data-structure into account – is not declared to be “compatible with the CHMP guideline”. At the end our gurus made an interesting statement about “Method C”:
      “This model […] will generally give wider confidence intervals
       than those produced by methods A and B.”
  Simply bizarre. Regulators suggesting a liberal/anti­con­ser­vative method. Once one of them at a conference made a strong statement: “We are interested in public health, not the profit of the pharmaceutical industry.” I’m not so sure…