PII-125 - QUANTITATIVE SYSTEMS TOXICOLOGY (QST) MODELING USING BIOLOGXSYM AND MECHANISTIC TOXICITY DATA FROM A BIOMIMETIC LIVER MICROPHYSIOLOGY SYSTEM PREDICTS BIOLOGICS-INDUCED LIVER INJURY (BILI) FOR MULTIPLE LARGE MOLECULES.

J. Beaudoin¹, L. Clemens¹, L. Vernetti², M. Miedel², M. Castiglione², L. Taylor², F. Huizar¹, C. Vallejo¹, C. Sandefur¹, M. Kelley¹, V. Lakhani¹, C. Battista¹, S. Siler¹, L. Shoda¹, B. Howell¹, K. Yang¹; ¹QSP Solutions, Simulations Plus Inc., Research Triangle Park, NC, United States, ²Drug Discovery Institute, University of Pittsburgh, Pittsburgh, PA, United States.

Background: While biologics offer promise in addressing a range of unmet medical needs, clinically observed BILI events are concerning for drug developers, health care providers and patients. QST modeling, combined with in vitro data, can improve understanding of BILI mechanisms and help predict hepatotoxicity in humans.

Methods: BIOLOGXsym is a QST model developed in MATLAB 2021a to simulate BILI liabilities by mathematically representing relevant hepatic biochemistry and physiology, as well as pathways and mechanisms unique to biologics (e.g., inhibition of IL-6 signaling by tocilizumab) 1. Mechanistic toxicity assay readouts from 10-day experiments with the Liver Acinus MicroPhysiology System (LAMPS) were evaluated after treatment with different biologics (GGF2, tocilizumab, ipilimumab, infliximab, bevacizumab 2,3) with and without BILI liabilities, and were used as BIOLOGXsym inputs to represent biologics-mediated hepatocyte stress signals (e.g., oxidative stress). Physiologically based pharmacokinetic models were developed in GastroPlus v9.8 to inform the exposure of these biologics in the hepatic interstitium at clinically relevant dosing protocols to drive the hepatotoxic effects in the BIOLOGXsym simulations.

Results: BIOLOGXsym simulations, combining exposure, LAMPS-informed toxicity mechanisms, and a virtual population of normal healthy volunteers (NHV SimPops, n=285) predicted plasma alanine aminotransferase (ALT) >3X upper limit of normal (ULN) for large molecules with clinical BILI liabilities: GGF2 (Sim: 0.7%, Data: 4.6% with ALT >3X ULN), tocilizumab (Sim: 6.7%, Data: 0.7-33.8% with ALT >1-5X ULN), ipilimumab (Sim: 15.1%, Data: 10.9% with ≥Grade 3), and infliximab (Sim: 0.7%, Data: 0.7% with ALT >3X ULN). For the negative control bevacizumab, no mechanistic LAMPS signal was observed, and no ALT elevations were simulated.

Conclusion: Overall, these simulation results show that QST modeling using BIOLOGXsym, while leveraging LAMPS experimentation, has the capacity to predict BILI liabilities for large molecules. For some biologics, low-level hepatocyte stress may sensitize the liver to T cell responses, which will be added to BIOLOGXsym subsequently.

Supported by NIH Award R44TR003535.

References: 1. Beaudoin et al. IJMS 2023; 2. Vernetti et al. SOT 2024; 3. Huizar et al. SOT 2024