Following oral administration to humans, amiodarone (AMD) is absorbed slowly and variably. The bioavailability of AMD is approximately 50% [1]. After a single dose, peak plasma concentrations are reached in 3 to 7 h. The relationships between plasma concentrations and the pharmacodynamic effect are very complex. It can appear in 2 to 3 days, but more often lasts one to three weeks, and even longer after dosing. AMD has a very large and variable volume of distribution, averaging about 60 L/kg due to its extensive accumulation in various compartments, especially adipose tissue and highly perfused organs [2]. A major metabolite of AMD, desethylamiodarone (dAMD), has been identified in humans. The main route of elimination is by hepatic excretion into the bile and significant enterohepatic circulation may occur.

In patients, AMD has been shown to have a biphasic elimination [3,4]. For dAMD, the mean half-life is 61 days. These data probably reflect an initial elimination of the drug from well-perfused tissues (half-life phase of 2.5 to 10 days) followed by a second phase with much slower elimination mainly from the fats.

The objectives of the studies were to compare the rate and degree of absorption, areas under the curves and variabilities for AMD administered in bioequivalence (BE) studies performed in the laboratory of the authors over the years.

Usual assessment of bioequivalence is performed in terms of mean plasma levels and pharmacokinetic parameters. As can be seen in Figure 1 (common representation in bioequivalence studies), mean plasma levels and mean maximum concentrations of the compared formulations were very close to each other. It should be noted that what is usually the maximum concentration, in the case of amiodarone, is a fluctuation, a “zone of maximum concentrations”, which is, as a rule, the consequence of the characteristic metabolic reactions of lipophilic drugs that lead to enterohepatic circulation, with a perturbation effect on the estimation of bioequivalence and with clinical implications [7,8,9,10,11,12].

However, beyond this first appearance, we must look for variability because it is very important for a particular patient if they are in the upper or lower part of the plasma level group, as the therapeutic and toxic effects are dependent on drug concentrations. Therefore, it is necessary to know well the main aspects of the variability in determining the benefit/risk ratio of the drug.

In the second stage, a characterization of the variability is added, using the standard deviations of the populations of concentrations corresponding to the different time points, as can be seen in Figure 2.

The variability of pharmacokinetics is an essential feature of the interaction between the drug and the living body, the rate and extent of release into the gastrointestinal tract, absorption, distribution, metabolism and excretion, ultimately determining the magnitude and rates of the therapeutic response and the balance between the efficacy and safety of an orally administered drug. In-depth evaluation of the PK/PD complex correlations involves a huge amount of research to overcome a great amount of difficulties. Under these conditions, the in-depth analysis of the results of bioequivalence studies, beyond comparing the bioavailability of two formulations, is not only an opportunity but also a moral duty. Last but not least, clinical trials are becoming more expensive and more and more complex, and regulatory restrictions are becoming more and more burning. Basic clinical research traditionally conducted by universities has been drastically reduced.

In BE studies, plasma concentrations are usually monitored at 15 to 30 time points, over two or more periods, usually on 24 subjects, representing a total of thousands of individual data. Factors that can be evaluated by chemical, mathematical and statistical methods are the pharmacokinetic parameters as well as their range of values and inter- and intrasubject variability.

In the case of the AMD lipophilic drug with a long half-life, there are particular difficulties in defining the maximum concentrations, but also the half-life and the elimination period. For the calculation of the total area, extrapolated from the area from zero to the last measuring point T, the formula shall apply: (2) AUC 0 − ∞ =   A U C 0 − T   +   A U C T − ∞   = A U C 0 − T +   C ( T ) / k e   =   A U C 0 − T   +   C ( T ) ln 2 t 1 / 2 so that the definition includes three relatively independent parameters, AUC_0–T , C(T), and t 1 / 2 ( k e ) .

In bioequivalence studies we are practically looking for the effects of the formulation and, in this context, we are less concerned about the elimination period. However, in the calculation of the total area under curve, we use t_1/2, whose uncertainty and variability are a real problem in the case of long half-life drugs. As a result of this variability, the confidence intervals for the mean ratio of AUC_0–∞ for the test and reference drugs are longer than the confidence intervals for the AUC_0–T ratio.

An FDA recommendation published in 2010 states: “For drugs that demonstrate low intrasubject variability in distribution and clearance, an AUC truncated at 72 h (AUC_0–72h) can be used in place of AUC_0–T or AUC_0–∞” [4]. However, the data collected and removed from the BE estimate remain very useful in research on the development of general or individual schemes of AMD therapy. Many unexplained facts remain in this regard, but a meta-analysis that includes a few more BE studies could lead to a partial understanding of the mechanisms of changes presented above.