The novel myeloperoxidase inhibitor verdiperstat was developed as a treatment for neuroinflammatory and neurodegenerative diseases. Phase 2 clinical studies had shown some promise for efficacy at the 600 mg BID dose; however, this is a large dose and verdiperstat had shown some in vitrosignals suggesting possible liver toxicity. Mild liver signals had also been observed during Phase 1 trials, though it was unclear whether these were drug-related or not. In order to provide an added layer of confidence in the liver safety of verdiperstat before proceeding to Phase 3, a computational prediction of verdiperstat liver safety was performed using DILIsymv8A, a quantitative systems toxicology (QST) model of liver safety.
A physiologically-based pharmacokinetic (PBPK) model of verdiperstat was constructed in GastroPlus9.8, and the estimates for the liver and plasma time course of verdiperstat were input into DILIsym. In vitro experiments measured the likelihood that verdiperstat would inhibit mitochondrial function, inhibit bile acid transporters, and generate reactive oxygen species (ROS). Predictions of liver verdiperstat exposure from the PBPK model and parameters derived from the in vitro experimental results were used as inputs into DILIsym. Two alternate sets of parameters were used as inputs in order to fully explore the sensitivity of model predictions within the potential range of the in vitro data. Verdiperstat dosing protocols up to 600 mg BID were simulated for up to 48 weeks using a simulated population (SimPops) in DILIsym.
In vitro experiments were conducted in cell vesicles (for bile acid transport) and HepG2 cells (for ROS and ETC inhibition). These experiments showed verdiperstat to be a mild inhibitor of the bile acid transporter MRP4 (Figure 1), a mild generator of ROS (Figure 2), and a mild inhibitor of the mitochondrial electron transport chain (ETC, Figure 3). For ROS and ETC inhibition, the intracellular concentration was measured by mass spectrometry.
Results from the in vitro experiments were used to calculate input parameters into DILIsymv8A, in the table below. OCR consumption was modeled in MITOsymv3B, a QST model of in vitro mitochondria, and translated into DILIsym; ROS generation was modeled in an in vitro-like parameterization in DILIsym (red lines in Figures 2 and 3). An alternate, conservative parameterization using an estimate of intracellular concentration as equal to the nominal concentration, which was suggested by the liver partition coefficient of 1 used in the PBPK model, was also developed; these parameters are also in the table below.
Verdiperstat was predicted to be safe, with only rare, mild liver enzyme increases as a potential possibility in very highly sensitive individuals. Subsequent Phase 3 clinical trials conducted after the conclusion of this modeling work found that ALT elevations in the verdiperstat treatment group were generally similar to those in the placebo group. This validates the DILIsym simulation results and demonstrates the power of QST modeling to predict the liver safety profile of novel therapeutics.
- BiohavenPharmaceuticals, Inc.
- The members of the DILI-sim and RENAsymInitiatives
Jeffrey L. Woodhead, Yeshi Gebremichael, Joyce Macwan, Irfan Qureshi, Richard Bertz, Victoria Wertz, Brett A. Howell
Society of Toxicology (SOT) 62nd Annual Meeting and ToxExpo, March 19-23, 2023, Nashville, Tennesee