Purpose: Sample size for bioequivalence study is calculated based on the intrasubject coefficient of variation (intra-CV%) of the pharmacokinetic (PK) parameters of drugs. In general, biopharmaceutics classification system (BCS) class 2 drugs with high permeability and low solubility are reported to have large variation in the sample size required for bioequivalence studies. In addition, poor oral absorption is a known risk factor to incur bioinequivalence of the test and reference drugs. Thus, we evaluated the factors affecting the intra-CV% by using the previously reported bioequivalence studies. Moreover, based on this, we predicted the intra-CV% of each drug according to the BCS class.
Methods: Data from the bioequivalence studies with 130 immediate release formulation drugs were used for the analysis. Intra-CV% was calculated as follows: 100×√(e^MSE-1), where MSE is the mean squared error. Geometric mean values for intra-CV% of the area under the concentration-time curve from time zero to the time of the last quantifiable concentration (AUClast) and the maximum concentration (Cmax) were estimated for each drug. Linear regression analyses were performed between log transformed geometric mean intra-CV% and factors including bioavailability, polar surface area, and the lowest solubility reported (solubility-lowest) for each drug according to the BCS class. R version 3.5.0 (R Foundation for Statistical Computing, Vienna, Austria) was used for statistical analyses.
Results: Among the 130 drugs, BCS class 2 (N=38) and BCS class 3 (N=29) drugs were included for linear regression. In terms of AUClast, log transformed bioavailability (log BA) was significant for BCS class 2 (log (intra-CV%) = 3.624–0.200×(log BA), R2=0.414), whereas both log BA and solubility-lowest were significant for BCS class 3 (log (intra-CV%) = 3.916–0.236×(log BA) –0.001×(solubility-lowest), R2=0.741). In case of Cmax, log BA was a significant factor affecting intra-CV% for both BCS class 2 (log (intra-CV%) = 3.903–0.173×(log BA), R2=0.339) and BCS class 3 (log (intra-CV%) = 3.914–0.151×(log BA), R2=0.266). Predicted intra-CV% were 15.0–51.3% (AUClast) and 22.4–65.0% (Cmax) for BCS class 2, whereas 7.5–55.7% (AUClast) and 25.0–53.8% (Cmax), for BCS class 3. Also, in the both class drugs of BCS 2 and 3, the proportion of total variability explained by the model of AUClast was larger than that explained by the model of Cmax.
Conclusion: This study showed that bioavailability is a significant factor affecting the intra-CV% of the PK parameters for both BCS class 2 and 3 drugs, and should be considered when calculating sample size for bioequivalence studies. In addition, solubility may have effects on the intra-CV% as well, when BCS class 3 drugs are used. Furthermore, we could successfully predict the intra-CV% of each drug by using the formula obtained from linear regression analyses.