Virtual Bioequivalence

The standard regulatory pathway for generic drug approval typically involves conducting bioequivalence (BE) studies. These studies aim to demonstrate that the generic product matches an already approved (reference) product in terms of the rate and extent of absorption. These aspects are commonly measured through pharmacokinetic parameters: peak concentration (C_max), representing the absorption rate, and the area under the concentration-time curve (AUC), indicating the extent of absorption. The standard methodology, known as average bioequivalence (ABE), requires that the 90% confidence interval for the ratio of geometric mean values (test versus reference) for both AUC and C_max falls within predefined regulatory acceptance limits, usually ranging from 80% to 125%.

However, when dealing with highly variable drugs (HVDs), traditional ABE approaches often face challenges. A drug is classified as highly variable if the within-subject variability (intra-subject coefficient of variation, C.V.) of its pharmacokinetic parameters, either AUC or C_max, exceeds 30%. Practically, this means that if the same person takes identical doses of a drug under similar conditions on two separate occasions, significant differences greater than 30% might still occur in measured drug concentrations. Due to this intrinsic variability, using conventional sample sizes in ABE studies may fail to establish bioequivalence even when formulations are genuinely similar. In fact, certain HVDs have struggled to demonstrate bioequivalence against themselves under typical ABE conditions.

To address this challenge, regulatory authorities increasingly recommend the Reference-Scaled Average Bioequivalence (RSABE) methodology. RSABE allows the acceptance criteria to be widened proportionally to the reference product's observed variability, thus accommodating the intrinsic variability inherent in HVDs. Without RSABE, bioequivalence studies for these products often need substantially larger sample sizes, driving up costs, exposing more subjects to potential risks, and ultimately restricting patient access to affordable generic medications.

The following equations describe the statistical analysis for virtual bioequivalence and bioequivalence assessment. You can find all the relevant equations and detailed explanations at the provided reference.

U.S. Food and Drug Administration (FDA). (2017). Guidance for Industry: Bioequivalence Studies With Pharmacokinetic Endpoints for Drugs Submitted Under an ANDA. Retrieved from https://www.fda.gov/media/87219/download.

Equation 6-7: Calculation of

where:

Equation 6-8: Determine the 95% upper confidence bound

Y_T  and Y_R are the means of the ln-transformed PK endpoint (AUC_0-t and AUC_0-inf and/or C_max) obtained from the BE study for the test and reference products respectively.

where:

scaled average BE limits.

= 0.25 Regulatory constant.

Acceptance criteria

AUC

If the within-subject variability (S_wr) is 0.294 or higher, then reference-scaled average bioequivalence (RSABE) is permitted and the 90% confidence interval (CI) acceptance criteria may be widened. However, the point estimate (or geometric mean ratio) must still fall within the 80–125% range. If S_wr is below 0.294, conventional average bioequivalence (ABE) methods should be employed

C_max

If the within-subject variability (S_wr) is 0.294 or above, reference-scaled average bioequivalence (RSABE) is allowed, and the 90% confidence interval acceptance range may be broadened. However, even with a widened acceptance criterion, the point estimate (or geometric mean ratio) must remain within 80–125%. If S_wr is less than 0.294, then the conventional average bioequivalence (ABE) methodology should be used.