Mathematically modeling CD8+ T cell-mediated drug induced liver injury (DILI): from ovalbumin to amodiaquine in mice

Conference: ACoP
Software: DILIsym®
Division: DILIsym Services



Idiosyncratic drug-induced liver injury (iDILI) is poorlynunderstood, but there is evidence to support that some iDILI events may be immune-mediated. Amodiaquine (AQ), a malaria medication, is thought to cause immune-mediated iDILI due its delay between treatment initiation and DILI onset and its ability to induce a more rapid injury upon drug rechallenge. 1 Previous work has been devoted to representing T cell-induced cytotoxicity in response to ovalbumin (OVA)-expressing hepatocytes in mice.2
The current work aims to translate the OVA representation to recapitulate AQmediated hepatotoxicity in mice.


An existing quantitative systems toxicology (QST) model (DILIsym®) was previously expanded to characterize CD8+ T cell responses to hepatocyte expressed OVA.2 This model has further been expanded to incorporateexposure of AQ via a physiologically-based pharmacokinetic (PBPK) representation which includes AQ metabolism to reactive metabolites (RMs).
AQ RMs induce hepatocellular ER stress3 as well as oxidative stress4. These mechanisms for intrinsic toxicity were evaluated for their potential to alter the hepatic microenvironment, making it more permissive for a CD8+ T cell response. The model representation of T cell interaction with hepatocytepresented antigen was also updated to incorporate the putative lower T cell receptor affinity of AQ compared with OVA and to represent a function of cytokine/co-stimulation levels which modulate T cell activity. These modifications, along with the pathway for T cell exhaustion already built into the OVA model, allow for representing interference in pathways implicated by CTLA4 and PD-1 compared with a wild-type mouse. Simulations aimed to reproduce CD8+ T cell-mediated DILI in a PD-1-/- mouse model (bred on a C57BL/6 background) exposed to AQ with anti-CTLA4 treatment.5,6


Simulations successfully captured differences in CD8+ T cell responses under different experimental conditions. No ALT elevations were predicted in simulated wild-type mice treated with AQ. Simulations designed to mimic inhibition of regulatory pathways, i.e., PD-1-/- mice treated with anti- CTLA4, predicted CD8+ T cell-mediated ALT elevations consistent with levels seen in experimental data.5,6. Variations in cytokine/co-stimulation levels were simulated to assess T cell response sensitivity and range of response.
These abstract results have been previously presented in part at Immunology2020, Honolulu, HI, 5/2020 and published in the conference proceedings as abstract 2395.


The results provided confidence in the QST model translation of OVA hepatotoxicity to AQ-induced DILI in mice. Results provided a basis for exploring AQ-induced DILI in humans using QST modeling with the goal of investigating immune-mediated iDILI risk for future compounds.

Presented at ACoP 11 Virtual Conference, Nov. 9-13, 2020

By Christina Battista, Zackary R. Kenz, Lisl K.M. Shoda