Abstracts

Rationale: During the past two decades concerns have been raised about the safety and efficacy of generic AEDs. The /- 20% range allowed for generic approval is often cited as being too broad. The decision process of how this range was selected may inform future study of generic AED bioequivalence. Methods: Extensive review of the literature and FDA documents obtained through a Freedom of Information Act request. Results: Confidence intervals (CIs) were first proposed for bioequivalence testing in 1972. In the early 1970 s the basic bioavailability assumption was enunciated: bioequivalent products must have similar rates and extent of availability of active ingredient in the blood. The pharmacokinetic measures were identified as the concentration maximum (Cmax) to assess rate and area under the curve (AUC) to assess availability. Hypothesis testing was deemed to be inadequate because two equivalent products could be statistically different when only a few percent separated the pharmacokinetic measures if a large enough sample size was used. During the 1970 s-80 s numerous authors proposed that bioequivalence would be established if the 90% CIs of the ratios of the means of the Cmax and AUCs of the generic and reference (usually brand) product were within a specified range. The example ranges provided were typically 15% or 20%. No randomized clinical trials were performed to determine the validity of any range. Many of the authors cautioned that the appropriate range may differ for different drugs and that the selected range should be based on individual drug clinical factors. In the late 1970 s, the /- 20% range was proposed by the FDA. At the time a different method, the 75/75 rule, was used to determine bioequivalence. In 1986 the FDA held a Bioequivalence Hearing. The hearing s conclusion was: differences of less than 20% in AUC and Cmax between products in normal subjects are unlikely to be clinically significant in patients. Clinical studies of effectiveness have difficulty detecting differences in dose of even 50-100%. Few drugs are given on a mg per kg basis to account for weight differences and few drugs have their dosage adjusted in actual clinical practice for factors that may affect blood level concentrations in individuals. Thus, the variability inherent in medical practice and biological variation may cause plasma levels to vary in individuals by much more than 20%. Because experts conclude that differences of less than 20% in mean AUC between brand name and generics are rarely unacceptable, FDA has established procedures to assure with high probability that the true mean AUC between brand and generic products do not differ by more than 20%. Subsequently, the FDA adopted the /- 20% range, modified to be that the log-transformed 90% CI of the generic product needed to be within 80-125% of the reference product in an average bioequivalence study. Conclusions: The selection of 80-125% for the allowable range of the 90% CI required to be met by generic drugs, including AEDs, was based on expert opinion and has never been studied in a randomized clinical trial.