Statistical Approaches to Establishing Bioequivalence

The FDA draft guidance on 'Statistical Approaches to Establishing Bioequivalence' discussed on the requirements for submitting bioavailability (BA) and bioequivalence (BE) data in investigational new drugs (INDs), new drug applications (NDAs), abbreviated new drug applications (ANDAs), and supplements; the definitions of BA and BE; and the types of in vitro and in vivo studies that are appropriate to measure BA and establish BE are set forth in part 320 (21 CFR part 320). 

This guidance provides recommendations to sponsors and applicants who intend to use equivalence criteria in analyzing in vivo or in vitro BE studies for INDs, NDAs, ANDAs, and supplements to these applications. This guidance discusses statistical approaches for BE comparisons and focuses on how to use these approaches both generally and in specific situations. When finalized, this guidance will replace the guidance for industry Statistical Approaches to Establishing Bioequivalence, which was issued in February 2001 (2001 guidance).

This guidance also provides recommendations on how to meet provisions of part 320 for all drug products.

The assessment of BE involves comparison between a test (T) and reference (R) drug product, where T and R can vary depending on the comparison to be performed (e.g., to-be-marketed formulation versus clinical trial formulation, generic drug versus RLD, originally approved formulation versus post-approval formulation changes).

Although BA and BE are closely related, BE comparisons normally rely on -

(1) a criterion,

(2) a confidence interval for the criterion,

(3) a predetermined BE limit.

BE comparisons could also be used in certain pharmaceutical product line extensions, such as additional strengths, new dosage forms (e.g., changes from immediate release to extended release), and new routes of administration. In these contexts, the approaches described in this guidance can be used to determine BE.

Study Design

1. Experimental Design

a. Nonreplicated designs

A conventional nonreplicated design, such as the standard two-formulation, two-period, two sequence crossover design, can be used to generate data when an average or population approach is chosen for BE comparisons. Under certain circumstances, such as products with apparent, long half-lives where crossover studies are impractical, parallel designs can be used.

b. Replicated crossover designs

Replicated crossover designs can be used irrespective of which BE approach is selected to establish BE, although they are not necessary when an average or population BE approach is used. When a reference-scaled BE approach is used, replicated crossover designs are critical to allow estimation of within-subject variances for the R (and T if a fully replicated study is used) measures. In particular, the following four-period, two-sequence, two-formulation design is recommended for fully replicated BE studies.

For this design, the same lots of the T and R formulations should be used for the replicated administration. Each period should be separated by an adequate washout period.

Other fully replicated crossover designs are also possible. For example, a three-period design, as shown below, could be used. A fully replicated design can estimate the subject-by-formulation interaction variance components. 

The following three-period, three-sequence, two-formulation, partially replicated design can also be used for assessing reference-scaled BE, though it cannot fully estimate the subject-by130 formulation interaction variance component (as a fully replicated design can). 

A greater number of subjects would be needed for the three-period designs compared to the recommended four-period design to achieve the same statistical power to conclude BE.

c. Adaptive design

An adaptive design is a clinical trial design that allows for prospectively planned modifications to one or more aspects of the design based on accumulating data from subjects in the trial. An adaptive design can be a group sequential design, or other design with one or more adaptive features.

For example, Potvin’s methods (Potvin et al. 2008, Xu et al. 2016) are a combination of a group sequential design and an adaptive design with sample size re-estimation.

Adaptive design can provide ethical advantages and statistical efficiency. When appropriately implemented, adaptive designs can reduce resources used, decrease time to study completion, and increase the chance of study success, especially when the prior information needed for the study design is limited. 

However, use of adaptive designs can also have limitations. For example, adaptive designs may call for certain statistical methods to avoid increasing the chance of erroneous conclusions and introducing bias in estimates and for complex adaptive designs, such methods may not be readily available. The decision to use or not use an adaptive design is at the applicant’s discretion.

d. Design with sparse sampling

For certain generic products, a sparse BE design is used, where the sampling for each subject is done at a single or very limited number of time points rather than the number needed to get a full concentration profile. For example, some ophthalmic products are studied using a sparse BE design, where only a single sample is collected from a single eye of each subject, at one assigned sampling time point for that subject. 

More generally, a sparse BE study design can be a parallel design where each subject should receive only one treatment, T or R, but not both. Alternatively, a crossover sparse study design can be used where each subject receives both test and reference treatments (e.g., in subjects undergoing indicated cataract surgery for both eyes).

2. Sample Size Determination 

It is an applicant’s responsibility to design an adequately powered BE study for the proposed study. We recommend that applicants enroll enough subjects to power the study at a level of 0.8 or higher, for a BE test to be carried out with a type 1 error rate of 0.05.

When determining the sample size, rates of attrition and noncompliance (e.g., protocol violation) should be taken into consideration. Enough subjects should be recruited, randomized, and dosed at the beginning of the study to ensure that the desired number of evaluable subjects will be available for analysis. All eligible subjects who were dosed should be  included in the analysis.

The number of subjects to be included in a study should be based on an appropriate sample size calculation for the proposed study design. For example, the standard 2×2 cross-over study will use a particular calculation while studies with a different design or set of endpoints will use  different calculations. 

In general, for PK BE or in vitro BE studies, sample size calculation should be based on BE metrics (e.g., AUC, Cmax) after log-transformation; for comparative clinical endpoint BE studies, sample size calculation should be based on the un-transformed comparative clinical endpoints unless otherwise noted in the relevant FDA product-specific guidance (PSG). The number of evaluable subjects in a PK BE study should not be less than 12. For highly variable drug products, a minimum of 24 subjects are recommended for BE assessment.

This guidance also discuss about -

  • Data Preparation
  • Statistical Models
  • Specific Situations
    • In Vitro Bioequivalence and Population Bioequivalence
    • Statistical Methods for Narrow Therapeutic Index and Highly Variable Drug Products
    • Comparative Clinical Endpoint Bioequivalence Studies
    • Studies in Multiple Groups
    • Bioequivalence Statistics for Adhesion and Irritation Studies

Post a Comment

Previous Post Next Post