Sanford Bolton

Bioequivalence studies for levothyroxine

The Food and Drug Administration (FDA) Guidance for Bioavailability and Bioequivalence Studies for Levothyroxine has been challenged by companies that manufacture brand-name products. Their contention is that the current guidance does not adequately address the endogenous background levels of the drug, and that the ratios of the PK parameters, a basis for approval of equivalence, are not assessed correctly. In particular, they conclude that products that have a potency differing by 12.5% cannot be differentiated using the present guideline and criteria for acceptance of bioequivalence. They claim that such a difference can be a public health hazard because of the perception among practitioners that levothyroxine is a narrow therapeutic index drug. This article describes the procedure recommended in the current Guidance for Levothyroxine and demonstrates that the methods recommended are adequate and will accept products that are therapeutically equivalent. To date, no generic product accepted as equivalent using FDA Guidances has been shown to result in a safety and efficacy profile different from its brand counterpart.

Generic Warfarin: Implications for Patient Care—Another View

Late in 1996, a submission to the Food and Drug Administration (FDA) of a generic warfarin product by Barr Laboratories engendered a controversy concerning substitution or switchability of narrow therapeutic range (NTI) drug products. A recent article in this journal suggested that pharmacists should be concerned about Substitution.’ The article strongly implied that an FDA-approved generic version of warfarin would likely result in considerable harm to patients who would switch to i t from Coumadin. We believe that this article contains a number of incorrect conclusions that are based on inaccurate interpretation of data. The controversy revolves around the nature of the narrow therapeutic range of warfarin and the influence of a generic formulation containing the identical dose and chemical composition of the active ingredient. Since the substitution of generic drugs with a narrow therapeutic index is currently being widely debated, we believe it is important to provide a more complete and accurate discussion and interpretation of the issues. The confusion surrounding warfarin stems from the fact that the tablet itself is not a variable product, but that the pharmacodynamic response is variable. The physicochemical and pharmacokinetic properties of warfarin sodium are predictive of a drug with extremely small variability Both generic and brand products are very uniform and show rapid dissolution. This

Experimental Design in Clinical Trials

Abstract The design and analysis of clinical trials is fertile soil for statistical applications. The use of sound statistical principles in this area is particularly important because of close FDA involvement, in addition to crucial public health issues which are consequences of actions based on the outcomes of clinical experiments. Principles and procedures of experimental design, particularly as applied to clinical studies, are presented. Relatively few different experimental designs are predominantly used in controlled clinical studies. In this chapter we discuss several of these important designs and their applications.

Independence and Statistical Inference in Clinical Trial Designs: A Tutorial Review

The requirements for statistical approaches to the design, analysis, and interpretation of experimental data are now accepted by the scientific community. This is of particular importance in medical studies where public health consequences are of concern. Investigators in the clinical sciences should be cognizant of statistical principles in general, but should always be wary of the pursuing their own analyses and engage statisticians for data analysis whenever possible. Examples of circumstances that require statistical evaluation not found in textbooks and not always obvious to the lay person are pervasive. Incorrect statistical evaluation and analyses in such situations will result in erroneous and potentially serious misleading interpretation of clinical data. Although a statistician may not be responsible for any misinterpretations in such unfortunate circumstances, the quote often cited about statisticians and “damned liars” may appear to be more truth than fable. This article is a tutorial review and describes a common misuse of clinical data resulting in an apparently large sample size derived from a small number of patients. This mistake is a consequence of ignoring the dependency of results, treating multiple observations from a single patient as independent observations.

Statistical Considerations: Alternate Designs and Approaches for Bioequivalence Assessments

PARALLEL DESIGN IN BIOEQUIVALENCE STUDIES The great majority of bioequivalence (BE) studies measure drug concentrations in body fluids, such that products can be compared within an individual using crossover designs. In some rare circumstances, this approach is either not possible or impractical. For example, drugs with long half-lives may not be amenable to a crossover design or studies where a clinical endpoint is required in patients because of insufficient blood concentrations. In these cases a parallel design may be used. In parallel designs comparative products are not given to the same patient. Patients are randomly assigned to one of the products under investigation. In this discussion, examples are used where two products are compared to each other, a Test and Reference product. Typically, a random device is used to assign product to patients as they enter the study, with an aim of having equal numbers of patients in each product group. For a BE study, it would be expected that patients would all be entered together, with each patient assigned a unique number. If more patients are needed than can be accommodated at one site, a multi-center study may be necessary. Randomization schemes for parallel studies have been described in the literature (1). Note that for these designs, the number of observations in each group does not need to be identical; drop-outs do not invalidate any of the remaining data. Endpoints in clinical studies can be “continuous” data or discrete. For example, the endpoint could be treadmill time to angina, or as for a local

An Alternative Derivation of the Distribution of the Individual Bioequivalence Metric

Abstract A method is proposed for estimating the probability density function (PDF) of the individual bioequivalence metric. We show that under the usual assumption of independence of the various random variables, the computations are quite feasible. A program to evaluate the integrals needed for the computations has been written. Experimentation with this program shows that the pass/fail decision based on the computed PDF compares favorably with that based on the method suggested in a recent FDA Guidance (Jan. 2001).