Bioequivalence and bioavailability testing are fundamental processes of a drug discovery workflow. Bioavailability studies assess the availability of the compound of interest at the site of drug action. On the other hand, bioequivalence analysis evaluates the therapeutic similarity between a test and a reference drug product. FDA recommends following the average bioequivalence (ABE) approach for bioequivalence clinical trials. Sponsors generally use a 2×2 crossover design with two drug formulations, two sequences, and two periods in bioequivalence clinical trials.
Pharmacokinetic properties such as Cmax and AUC are estimated during bioequivalence analysis. Bioequivalence studies are evaluated by calculating a confidence interval of 90% for the geometric least square mean ratio. Two drug formulations are considered bioequivalent when both confidence intervals are within 80-125%. However, for a highly variable drug, a large sample population must assess within-subject variability. But as regulatory bodies have raised concerns regarding using a large sample population, research studies suggest using linear mixed model simulations for bioequivalence assays.
Even when it alleviates specific issues related to large study samples, linear mixed models still face challenges, particularly while reflecting physiological characteristics and calculating sample sizes. Hence, the current article highlights a novel approach for evaluating the bioequivalence of new drug products.
An efficient model-based bioequivalence testing
Bioequivalence is a vital component of drug discovery steps, and hence they will need a robust clinical trial design. This bioequivalence approach is based on FDA guidelines. The current PK-based model employs 2×2, 3×3, and 2×4 crossover designs. Let us dive deep into the design of a model-based bioequivalence study.
As mentioned earlier, the study simulations are conducted in three crossover designs. A two-compartment model assesses drug concentration between the reference and test drug product. Moreover, the study incorporates PK parameters such as clearance, the volume of distribution of the central and peripheral compartment, bioavailability, and the absorption rate constant.
The model assumes lognormal distribution for random effects and sets the mean individual effects to zero. It calculates standard deviation and intra-subject variability using clopidogrel data. Moreover, the model-based approach incorporates typical bioavailability values between the two drug products. Besides, intra-subject variability is calculated based on intra-subject CV values and is set as ≥0.3 for assessing highly variable drug products.
Individual values for Vp, Vc, Q, CL, F, and Ka are derived through the PK-based model. Researchers obtain each simulation concentration for all three clinical trial designs. The model uses an equation to generate individual PK parameters. The equation considers the dose, elimination rate constant of the central compartment, and transfer rate constant between the peripheral and central compartments. Finally, the AUC or Cmax is calculated from the simulation models, and bioequivalence is investigated based on FDA methods. The simulation is continuously conducted 1000 times for each assay condition, and the statistical power is calculated for a 90% confidence interval for AUC or Cmax values.
A model-based bioequivalence testing is critical for the success of a generic drug product. Although it is more beneficial than linear mixed models, they can be complex and time-consuming. Hence, appropriate planning and execution will be necessary for model-based bioequivalence testing.