James Craig1, Natalie Morris2, Kevin McNally2, Clairissa Corpstein2, Krishna Chaitanya Telaprolu2, Pauline Bogdanovich2, David Turner2, Amin Rostami-Hodjegan2, Masoud Jamei2, Frederic Y. Bois2
1Certara Data Science Software, 2Certara Predictive Technologies
Objectives: Demonstrating bioequivalence (BE) for drug formulations with different critical quality attributes (CQA) in vivo can be a time and resource intensive process, however virtual bioequivalence (VBE) studies can help streamline these assessments. The Pirana-Simcyp VBE module was developed to provide a user-friendly framework for the organization and optimization of PBPK model-based VBE workflows [1-2]. Pirana interfaces with the Simcyp-R package to run PBPK simulations [3] and provides bespoke scripts for data analysis and visualisation. To demonstrate its applications and functionality, we utilize the Pirana VBE module to implement a case study on 3-month long-acting injectable suspension formulations of paliperidone palmitate (PP3M), in which the particle size distribution (PSD) initial mean radius is taken as the CQA [4-5]. This work aimed to (1) showcase a systematic approach for conducting VBE analyses using the Pirana–Simcyp integration, and (2) illustrate how the mean drug particle size influences bioequivalence outcomes for PP3M in virtual populations. Methods: A physiologically based pharmacokinetic (PBPK) model of a PP3M reference formulation was defined in a Simcyp workspace, with an initial mean particle radius of 4.6357 µm. Virtual population simulations were conducted in Simcyp for reference and test formulations using a range of PSD mean radii (1.80–7.50 µm), with up to 100,000 subjects per simulation. Using Pirana’s Simcyp-VBE Shiny application and integrated R script capabilities, we performed the following key workflow steps: 1.Sensitivity Analysis: Population representative simulations were used to explore the impact of varying PSD means on plasma concentration profiles. 2.Full Virtual Trials: Large-scale simulations were performed for each mean radius value, generating individual time–concentration profiles. 3.Data Processing/QC: Automated scripts in Pirana aggregated and quality-checked the outputs (e.g., negative values, uniform time points). 4.Power Analysis: Combined test/reference datasets were analyzed with a Monte Carlo bootstrap resampling algorithm to assess the probability of determining bioequivalence as a function of trial size. This allowed estimation of the trial sizes required to achieve over 80% power. 5.Type I Error: Using regression-based sensitivity analyses, PSD mean values predicted to result in population geometric mean ratios (GMRs) of 0.80 and 1.25 were identified. Type I error was then determined by simulating VBE trials at those boundary values. 6.Safe Space: The range of initial mean particle sizes likely to result in successful VBE trials with =80% power was mapped for a given trial size. Results: All simulations successfully generated time–concentration datasets for test and reference formulations. Power analyses suggested that when provided a perfectly bioequivalent test formulation, approximately 160 subjects per arm were required to reach 80% power. A deviation in the test particle size of only 20% from the reference doubled the sample size required to maintain 80% power. Regression-based sensitivity analyses demonstrated a linear relationship between PSD mean and the GMR of C_max. Type I error analyses identified that particle sizes predicted to yield a GMR of 0.80 and 1.25 had approximately a 5% chance of erroneously being declared bioequivalent, indicating that the VBE assessment appropriately controlled type I error. Safe space evaluations assuming a parallel trial size of 200 patients per arm confirmed that PSD mean values within ±10% of the reference remained within the targeted bioequivalence range with a statistical power of over 80%. Conclusions: This case study illustrates how a streamlined VBE workflow—combining Simcyp’s PBPK modeling and Pirana’s automated data handling—can be used to efficiently simulate trial outcomes for alternative formulations of paliperidone palmitate. Pirana provides user-friendly data visualization tools and a set of template-based R scripts to support the VBE workflow, with all scripts being customizable to support specific analysis objectives. These results highlight how in silico analyses can guide formulation development and trial design decisions by estimating trial feasibility and defining safe spaces for CQAs, without extensive in vivo testing. The Pirana-Simcyp VBE integration demonstrates how an analyst can perform robust VBE studies via an automated, reproducible, and extensible R-based workflow.
1. Chen R. et al. (2025). Pirana and integrated PMX tools. CPT Pharmacometrics Syst Pharmacol. In review. 2. Craig J. (2024). Certara.SimcypVBE: Shiny GUI for Simcyp VBE Workflow (Version 1.0.0) [R package]. Available at: https://certara.github.io/R-SimcypVBE/. 3. Vinden B. (2024). Simcyp: A Package to Support R Script Interactions with the Simcyp Simulator (Version 24.0.18) [R package]. Available at: https://members.certara.co.uk/ 4. Morris NM et al. (2025). Virtual bioequivalence of long-acting injectable suspensions using PBPK modeling: Part 1. Power analysis. In review. 5. Morris NM et al. (2025). Virtual bioequivalence of long-acting injectable suspensions using PBPK modeling: Part 2. Type 1 error and safe-space analyses. In review.
Reference: PAGE 33 (2025) Abstr 11584 [www.page-meeting.org/?abstract=11584]
Poster: Software Demonstration