The primary concern in bioequivalence assessment is to limit the risk of a false declaration of equivalence. Statistical analysis of the bioequivalence trial should demonstrate that a clinically significant difference in bioavailability between the multisource product and the comparator product is unlikely. The statistical procedures should be specified in the protocol before the data collection starts.
The statistical method for testing pharmacokinetic bioequivalence is based upon the determination of the 90% confidence interval around the ratio of the log-transformed population means (multisource/comparator) for the pharmacokinetic parameters under consideration and by carrying out two one-sided tests at the 5% level of significance.
All concentration-dependent pharmacokinetic parameters (e.g. AUC and Cmax) should be log-transformed using either common logarithms to the base 10 or natural logarithms. The choice of common or natural logs should be consistent and should be stated in the study report.
Logarithmically transformed, concentration-dependent pharmacokinetic parameters should be analysed using analysis of variance (ANOVA). Usually the ANOVA model includes the formulation, period, sequence or carry-over and subject factors.
Parametric methods, i.e. those based on normal distribution theory, are recommended for the analysis of log-transformed bioequivalence measures. The antilog of the confidence limits obtained constitute the 90% confidence interval for the ratio of the geometric means between the multisource and comparator products. The same procedure should be used for analysing parameters from steady state trials or cumulative urinary recovery, if required.
For tmax descriptive statistics should be given. If tmax is to be subjected to a statistical analysis this should be based on non-parametric methods and should be applied to untransformed data. A sufficient number of samples around predicted maximal concentrations should have been taken to improve the accuracy of the tmax estimate. For parameters describing the elimination phase (T1/2) only descriptive statistics should be given.
Methods for identifying and handling of possible outlier data should be specified in the protocol. Medical or pharmacokinetic explanations for such observations should be sought and discussed. As outliers may be indicative of product failure, post hoc deletion of outlier values is generally discouraged.
An approach to dealing with data containing outliers is to apply distribution-free (non-parametric), statistical methods.
If the distribution of log-transformed data is not normal, non-parametric statistical methods can be considered. The justification of the intent to use nonparametric statistical methods should be included a priori in the protocol.
Area under the curve-ratio
The 90% confidence interval for this measure of relative bioavailability should lie within a bioequivalence range of 0.80–1.25. If the therapeutic range is particularly narrow, the acceptance range may need to be reduced based on clinical justification. A larger acceptance range may be acceptable in exceptional cases if justified clinically.
In general the acceptance limit 0.80–1.25 should be applied to the Cmax-ratio. However, this measure of relative bioavailability is inherently more variable than, for example, the AUC-ratio, and in certain cases a wider acceptance range (e.g. 0.75–1.33) may be acceptable. The range used must be defined prospectively and should be justified, taking into account safety and efficacy considerations. In exceptional cases, a simple requirement for the point estimate to fall within bioequivalence limits of 0.80–1.25 may be acceptable with appropriate justification in terms of safety and efficacy.
Statistical evaluation of tmax makes sense only if there is a clinically relevant claim for rapid onset of action or concerns about adverse effects. The nonparametric 90% confidence interval for this measure of relative bioavailability should lie within a clinically relevant range.
For other pharmacokinetic parameters the same considerations as outlined above apply.
Number of subjects
The number of subjects required for a sound pharmacokinetic bioequivalence study is determined by:
• The error variance (coefficient of variation) associated with the primary parameters to be studied, as estimated from a pilot experiment, from previous studies or from published data;
• The significance level desired (5%);
• The statistical power desired;
• The mean deviation from the reference product compatible with bioequivalence and with safety and efficacy;
The need for the 90% confidence interval around the geometric mean ratio to be within 80–125% bioequivalence limits for log transformed data. Pilot study should have reasonable size (n=16) for sample size planning for pivotal study and also selecting best of several formulations.
The number of subjects to be recruited for the study should be estimated by considering the standards that must be met. It should be calculated by appropriate methods. The number of subjects recruited should always be justified by the sample-size calculation provided in the study protocol. A minimum of 12 subjects is required.
Drop-outs and withdrawals
Sponsors should select a sufficient number of study subjects to allow for possible drop-outs or withdrawals. Because replacement of subjects during the study could complicate the statistical model and analysis, drop-outs generally should not be replaced. Reasons for withdrawal (e.g. adverse drug reaction or personal reasons) must be reported.
Dr. Manoj Karwa
Did you get here from a link from a friend, or twitter? This lesson is 9 of 10 parts Bio-equivalence Free e-course. To get more information about it and sign up Click here