Highly Variable Drugs

messages in brief | 14/02/2013

If there are major differences in the measurement of pharmacokinetic parameters, such as the AUC or Cmax, after repeated administration of the identical drug, it is reasonable to suspect that the specific product or the active ingredient itself could be a "highly variable drug" (HVD). By definition, it is justified to speak of an HVD as soon as the intra-individual variability (CV) of these parameters is greater than 30 percent.


The problem that arises for applicants of HVDs is that a high CV value requires an unequally higher number of subjects in a bioequivalence study (BE study) in order to be able to show the required equality with sufficient certainty.


In the literature, therefore, failure rates of 60 to 80 percent are found in corresponding BE studies for HVDs with a conventional number of subjects (approx. twelve to 24 subjects). Whereas statistically speaking, e.g. a BE study with 88 subjects (which, mind you, is more than seven times the minimum number of subjects required by the guideline) shows a nearly 100 percent probability of equality at a CV value of 15 percent, this will only succeed in 45 percent of all cases at a CV value of 30 percent.


This dilemma sometimes leads to the point that even completely identical HVD products have massive problems to prove bioequivalence - even with themselves. This is described, among others, in studies with the active ingredients verapamil, chlorpromazine and esomeprazole.


Possible causes for HVD behavior of active ingredients are predominantly due to a pronounced first-pass metabolism or also to a particularly low bioavailability of active ingredients. In rarer cases, galenics play a role. In order to avoid an ethically hardly justifiable increase in the number of test subjects to unrealistic levels for these products, it was decided after intensive EU-wide discussions to approve facilitations for HVDs under certain conditions [1].


Thus, for an HVD, depending on its actual CV value, a widening of the conventional allowed confidence interval limits from originally 0.80 to 1.25 (or 80 to 125 percent can be applied. The extent of the permitted widening depends on the respective level of the CV and starts at a CV value of 30 percent and continues arithmetically up to a value of 50 percent. For values above 50 percent, the permitted confidence interval upper limit is always 69.84 to 143.19 percent.


With scaled-average-bioequivalence, an associated confidence interval (CI) limit can be calculated for each CV value. For this purpose, the formula [U, L] = exp [±k-sWR] is used, where "U" represents the upper and "L" the lower acceptance limit of the confidence interval, "k" the so-called regulatory constant, which has a value of 0.760, and "sWR" the intra-individual CV, i.e. the standard deviation of the intra-individual, log-transformed pharmacokinetic data from the study.


However, the possibility to widen the limits of the confidence interval exists exclusively for the Cmax and is not provided for the AUC. However, it is always a prerequisite that it has been demonstrated beyond doubt in the BE study that the CV is greater than 30 percent. Only the value of the reference product is to be used as the undoubted CV, but not that of the test product or the generic product. However, a conventional two-way cross-over study design is not sufficient to prove "CV > 30 percent" beyond doubt. Rather, a replicate design is required, i.e. the reference product must be administered at least twice. The minimum requirement is therefore a semi-replicate design with three periods (e.g. RTR, RRT, TRR) in order to calculate the intra-individual variation of the originator.


Restricting the extended confidence intervals, there is an additional requirement that the calculated mean value of the Cmax, i.e. the geometric mean ratio (GMR) should lie within the conventional limits of 80 to 125 percent. This "double end point approach" thus helps, with all the flexibility granted to HVDs, to reliably avoid excessively large outliers.


If a marketing authorization is intended with a substance that could be an HVD, it is important for the applicant to consider in advance whether to conventionally accept the increased risk of failure of the BE study, to increase the number of subjects accordingly, or alternatively to consider a replicate design in his study planning. In case of doubt, both the national authority (BASG/AGES Medizinmarktaufsicht) and the European Medicines Agency (EMA) can be consulted for support by means of a scientific advice procedure.


[1] EMA: CPMP/EWP/QWP/1401/98 Rev. 1, Guideline on the Investigation on Bioequivalence.


Table: Examples for maximum limits of the confidence interval for Cmax, depending on the extent of intra-individual CV


within-subject CV (%)Lower LimitUpper Limit
> 5069.84143.19




Institute Admission & Lifecycle ManagementDr. Christoph Baumgärtel Tel.: +43 (0) 50 555-36540 E-mail: c hristoph.baumgaertel@ages.at


Further inquiry note