Population Model Development
The population model fitting in the observed healthy volunteer study data revealed that a two-compartment pharmacokinetic model adequately described the pharmacokinetics of 6MP after numerous iterative steps. The compartment-based pharmacokinetic model fit indicated significantly different absorption rates for the Test and the Reference product. Therefore, the cumulative fraction dissolved from the in-vitro dissolution study was fitted to a first-order exponential function to assess the rate of dissolution specific to the products which were then incorporated as fixed parameters specific to the formulation in the model. The final PK model described the pharmacokinetics of Mercaptopurine PFOS 10mg/mL after oral dosing with a dissolution and transit compartment followed by two compartmental pharmacokinetics. The PK model parameterized the volume of distribution of central and peripheral compartments (denoted as V1 and V2 respectively), clearance parameters from the central as well as inter-compartment (cl and Q respectively), and a first-order absorption rate (ka). Figure 3 describes the schematic representation of the PK model
The final model parameters for the PK model fit into the 51 subject clinical data for Test and Reference products along with their inter-individual variabilities are illustrated in Table 3. Diagnostic plots for the population PK fit for Reference and test product indicate an acceptable fit of the model. The diagnostic plots of the model fit for the reference and test products are presented in supplementary figures 2 and 3 respectively.
Replication of BE results for Test and Reference product
The model fitted population parameters were used to simulate the PK of the Test and Reference products in 51 virtual subjects. The model simulation results matched the observed clinical data closely. Figure 4 shows the observed vs simulated concentration-time profiles for the test and reference. The Cmax and AUC values calculated from the actual data and simulated data were comparable with a ratio ranging from 0.78 to 1.02 (Table 4). The BE assessment results for the model simulated data matched the results observed using clinical data. The 90 % CI of the geometric means were comparable (Supplementary table 3).