Bioequivalence studies are designed to compare the in vivo performance of a test multisource pharmaceutical product with that of a reference pharmaceutical product. A common design for a bioequivalence study involves the administration of the test and reference products on two occasions to volunteer subjects, the second administration being separated from the first by a wash-out period of duration such as to ensure that the drug given in the first treatment is entirely eliminated before the second treatment is administered. Just before administration and for a suitable period afterwards, blood and/or urine samples are collected and assayed for the concentration of the drug substance and/or one or more metabolites. The rise and fall of these concentrations over time in each subject in the study provide an indication of how the drug substance is released from the test and reference products and absorbed into the body. To allow comparisons between the two products, these blood (including plasma or serum) and/or urine concentration - time curves are used to calculate certain bioequivalence metrics of interest. Commonly used metrics include the area under the blood (plasma or serum) concentration - time curve (AUC) and the peak concentration. These are calculated for each subject in the study and the resulting values compared statistically. Details of the general approach are given below.
Selection of subjects
The subject population for bioequivalence studies should be as homogeneous as possible; studies should therefore generally be performed with healthy volunteers so that variability, other than in the pharmaceutical products concerned, is reduced. Clear criteria for inclusion/exclusion should be established. If possible, subjects should be of both sexes; however, the risk to women will need to be considered on an individual basis and, if necessary, they should be warned of any possible dangers to the fetus if they should become pregnant. They should normally be in the age range 18-55 years and of weight within the normal range according to accepted life tables. Subjects should preferably be non-smokers and without a history of alcohol or drug abuse. If smokers are included, they should be identified as such. Volunteers should be screened for suitability by means of standard laboratory tests, a medical history, and a physical examination. If necessary, special medical investigations may be carried out before and during studies, depending on the pharmacology of the drug being investigated.
If the aim of the bioequivalence study is to address specific questions (e.g. bioequivalence in a special population), the selection criteria will have to be adjusted accordingly.
Phenotyping and/or genotyping of subjects may be considered for safety reasons.
Patients versus healthy volunteers
If the active substance is known to have adverse effects and the pharmacological effects or risks are considered unacceptable for healthy volunteers, it may be necessary to use patients under treatment instead. This alternative should be explained by the sponsor.
Monitoring the health of subjects during the study
During the study, the health of volunteers should be monitored so that the onset of side-effects, toxicity, or any intercurrent disease may be recorded, and appropriate measures taken. Health monitoring before, during and after the study must be carried out under the supervision of a qualified medical practitioner licensed in the jurisdiction in which the study is conducted.
General study design
The study should be designed so that the test conditions are such as to reduce intra- and intersubject variability and avoid biased results. Standardization of exercise, diet, fluid intake and posture, and restriction of the intake of alcohol, caffeine, certain fruit juices, and drugs other than that being studied in the period before and during the study are important in order to minimize the variability of all the factors involved except that of the pharmaceutical product(s) being tested.
A cross-over design with randomized allocation of volunteers to each leg is the first choice for bioequivalence studies. Study design should, however, depend on the type of drug, and other designs may be more appropriate in certain cases, e.g. with highly variable drugs and those with a long half-life. In cross-over studies, a wash-out period between the administration of the test product and that of the reference product of more than five times the half-life of the dominant drug is usual, but special consideration will need to be given to extending this period if active metabolites with longer half-lives are produced, and also under certain other circumstances.
The administration of the test product should be standardized, i.e. the rime of day for ingestion and the volume of fluid (150 ml is usual) should be specified. Test products are usually administered in the fasting state.
Parameters to be assessed
In bioavailability studies, the shape of, and the area under, the plasma concentration curve, or the profile of cumulative renal excretion and excretion rate are commonly used to assess the extent and rate of absorption. Sampling points or periods should be chosen such that the time-concentration profile is adequately defined so as to allow the calculation of relevant parameters. From the primary results, the bioavailability parameters desired, e.g. AUC∞, AUCt, Cmax, tmax, Ae∞, Aet, dAe/dt, or any other necessary parameters, are derived (see Appendix 2). The method of calculating AUC-values should be specified. AUC∞ and Cmax are considered to be the most useful parameters for the assessment of bioequivalence. For urine excretion data, the corresponding parameters are Ae∞ and dAe/dtmax. For additional information, t½ and MRT can be calculated, and for steady-state studies, AUCτ, and the per cent peak - trough fluctuation. The exclusive use of modelled parameters is not recommended unless the pharmacokinetic model has been validated for the active substance and the products.
Additional considerations for complicated drugs
For drugs which would cause unacceptable pharmacological effects (e.g. serious adverse events) in volunteers or where the drug is toxic or particularly potent or the trial necessitates a high dose, cross-over or parallel-group studies in patients may be required.
Drugs with long half-lives may require a parallel design or the use of truncated area under curve (AUCt) data or a multidose study. The truncated area should cover the absorption phase.
For drugs for which the rate of input into the systemic circulation is important, more samples may have to be collected around the time tmax.
Multidose studies may be helpful in assessing bioequivalence for:
- drugs with non-linear kinetics (including those with saturable plasma protein binding);
- drugs for which the assay sensitivity is too low to cover a large enough portion of AUC∞;
- drug substance combinations, if the ratio of the plasma concentration of the individual drug substances is important;
- controlled-release dosage forms;
- highly variable drugs.
Number of subjects
The number of subjects required for a sound bioequivalence study is determined by the error variance associated with the primary parameters to be studied (as estimated from a pilot experiment, from previous studies or from published data), by the significance level desired, and by the deviation from the reference product compatible with bioequivalence, safety and efficacy. It should be calculated by appropriate methods (see below) and should not normally be smaller than 12. In most studies, 18-24 subjects will be needed (7-9). The number of subjects recruited should always be justified.
The products (samples) used in bioequivalence studies for registration purposes should be identical to the projected commercial pharmaceutical product. For this reason, not only the composition and quality characteristics (including stability) but also the methods of manufacture should be those to be used in future routine production runs.
Samples should ideally be taken from industrial-scale batches. When this is not feasible, pilot- or small-scale production batches may be used provided that they are not less than one-tenth (10%) of the size of the expected full-scale production batches.
It is recommended that the potency and in vitro dissolution characteristics of the test and reference pharmaceutical products should be ascertained before an equivalence study is performed. The content of active drug substance(s) in the two products should not differ by more than ±5%. If the potency of the reference material deviates by more than 5% from that corresponding to the declared content of 100%, this difference may be used subsequently to dose-normalize certain bioavailability metrics in order to facilitate comparisons between the test and reference pharmaceutical products.
Studies of metabolites
The use of metabolite data in bioequivalence studies requires careful consideration. The evaluation of bioequivalence will generally be based on the measured concentrations of the pharmacologically active drug substance and its active metabolite(s), if present. If it is impossible to measure the concentration of the active drug substance, that of a major biotransformation product may be measured instead, while measurement of the concentration of such a product is essential if the substance studied is a prodrug. If urinary excretion (rate) is measured, the product determined should represent a major fraction of the dose. Although measurement of a major active metabolite is usually acceptable, that of an inactive metabolite can only rarely be justified.
Measurement of individual isomers for chiral drug substance products
A non-stereoselective assay is currently acceptable for bioequivalence studies. Under certain circumstances, however, assays that distinguish between the enantiomers of a chiral drug substance may be appropriate.
Validation of analytical procedures
All analytical procedures must be well characterized, fully validated and documented, and satisfy the relevant requirements as to specificity, accuracy, sensitivity and precision. Knowledge of the stability of the active substance and/or biotransformation product in the sample material is a prerequisite for obtaining reliable results (10). It should be noted that:
- validation comprises both before-study and within-study phases;
- validation must cover the intended use of the assay;
- the calibration range must be appropriate to the study samples;
- if an assay is to be used at different sites, it must be validated at each site and cross-site comparability established;
- an assay which is not in regular use requires sufficient revalidation to show that it is performed according to the original validated procedures; the revalidation study must be documented usually as an appendix to the study report;
- within a given study, the use of two or more methods to assay samples in the same matrix over a similar calibration range is strongly discouraged;
- if different studies are to be compared, the samples from these studies have been assayed by different methods, and the methods cover a similar concentration range and the same matrix, they should be cross-validated.
The results of validation should be reported.
Sufficient samples of each batch of the pharmaceutical products used in the studies, together with a record of their analyses and characteristics, must be kept for reference purposes under appropriate storage conditions as specified by national regulations. At the specific request of the competent authorities, these reserve samples may be handed over to them so that they can recheck the products.
Statistical analysis and acceptance criteria
The primary concern in bioequivalence assessment is to limit the risk (a) of a false declaration of equivalence to that which the regulatory authorities are willing to accept.
The statistical methods of choice at present are the two one-sided tests procedure (11) and the derivation of a parametric or non-parametric 100(1-2α)% confidence interval for the quotient µT/µR of the test and reference pharmaceutical products. The value of a is set at 5%, leading, in the parametric case, to the shortest (conventional) 90% confidence interval based on an analysis of variance or, in the non-parametric case, to the 90% confidence interval (12, 13).
The statistical procedures should be specified before data collection starts (see Appendix 3), and should lead to a decision scheme which is symmetrical with respect to the two formulations, i.e. it should lead to the same decision whether the new formulation is compared with the reference product or vice versa.
Concentration and concentration-related quantities e.g. AUC and Cmax, should be analysed after logarithmic transformation, but tmax will usually be analysed without such transformation.
For tmax, normally descriptive statistics should be given. If tmax is to be subjected to a statistical analysis, this should be based on non-parametric methods. Other parameters may also be evaluated by non-parametric methods, when descriptive statistics should be given that do not require specific distributional assumptions, e.g. medians instead of means.
The assumptions underlying the design or analysis should be addressed, and the possibility of differing variations in the formulations should be investigated. This covers the investigation of period effects, sequence or carry-over effects, and homogeneity of variance.
The impact of outlying observations on the conclusions should be reviewed. Medical or pharmacokinetic explanations for such observations should be sought.
For AUC, the 90% confidence interval should generally be within the acceptance range 80-125%. For drugs with a particularly narrow therapeutic range, the AUC acceptance range may need to be smaller; this should be justified clinically.
Cmax does not characterize the rate of absorption particularly well in many cases, but there is no consensus on any other concentration-based parameter which might be more suitable. The acceptance range for Cmax may be wider than that for AUC (see Appendix 3).
Reporting of results
The report on a bioequivalence study should give the complete documentation of its protocol, conduct and evaluation in compliance with the guidelines on good clinical practice (GCP) for trials on pharmaceutical products (5). The responsible investigator(s) should sign the respective section(s) of the report. The names and affiliations of the responsible investigator(s), the site of the study and the period of its execution should be stated. The names and batch numbers of the pharmaceutical products used in the study, as well as the composition(s) of the tests product(s), should also be given. The analytical validation report should be attached. The results of in vitro dissolution tests should be provided. In addition, the applicant for registration should submit a signed statement confirming that the test product is identical with the pharmaceutical product submitted.
All results should be clearly presented. The procedure for calculating the parameters used (e.g. AUC) from the raw data should be stated. Deletion of data should be justified. If results are calculated using pharmacokinetic models, the model and the computing procedure used should be justified. Individual plasma concentration - time curves should be drawn on a linear/linear scale, and may also be shown on a linear/log scale. All individual data and results should be given, including those for any subjects who have dropped out of the trial. Drop-out and withdrawal of subjects should be reported and accounted for. Test results on representative samples should be included.
The statistical report should be sufficiently detailed to enable the statistical analyses to be repeated, if necessary. If the statistical methods applied deviate from those specified in the trial protocol, the reasons for the deviations should be stated.