This study presents a workflow for virtual bioequivalence (VBE) assessment of 3-month paliperidone palmitate (PP) long-acting injectable (LAI) suspensions using a novel physiologically based pharmacokinetic (PBPK) model. The mechanistic absorption and PBPK model was implemented in the Simcyp® Simulator and calibrated against individual concentration-time profiles derived from a published and validated population pharmacokinetic model. The model was able to accurately simulate drug concentration profiles after PP administration. Across 1000 subjects, four model-predicted bioequivalence (BE) metrics, including Cmax,ss and AUCtau,ss, differed by at most 10% from validation data. The initial mean drug particle radius was assessed as the critical formulation attribute in the VBE analysis. We conducted extensive VBE simulations to evaluate the required sample size of parallel trial designs for a given statistical power. The statistical power of two-one-sided t tests (TOST) to declare BE was estimated from the passing rate of Monte Carlo simulated VBE trials. Power calculations using the validated model indicated that a VBE trial with a minimum of 160 subjects per arm is required to achieve at least 80% power for declaring BE when the formulations are identical in terms of mean particle size. If the mean drug particle radius between test and reference formulations differs by 20%, the required sample size for BE demonstration approximately doubles to maintain the same power. This suggests that particle size affects formulation variability. The power calculations demonstrated that BE assessments were very sensitive to formulation differences in drug particle radius and to the parametrization of the model. These findings emphasize the critical need for rigorous model validation to ensure reliable VBE assessments.
Morris et al. (Fri,) studied this question.