Mechanistic Modeling With Dilisym® Predicts Dose-Dependent Clinical Hepatotoxicity Of Amg 009 That Involves Bile Acid Transporter Inhibition

Conference: ACoP
Software: DILIsym®
Division: DILIsym Services

Abstract

Objectives: To predict the clinical hepatotoxicity of AMG 009 and species differences in AMG 009-mediated hepatotoxicity using DILIsym®, a mechanistic model of drug-induced liver injury

Methods: Inhibitory effects of AMG 009 for bile acid transporters were assessed using transporteroverexpressing vesicles and transfected cells. Hepatotoxicity responses to AMG 009 in humans and rats were simulated in DILIsym® using a PBPK model of AMG 009, bile acid homeostasis and toxicity submodels, 1,2 and in vitro bile acid transporter inhibition data. Previously constructed human and rat simulated populations (SimPopsTM) that incorporate variability in bile acid disposition and mitochondrial function were employed for population analyses.

Results: AMG 009 was a mixed inhibitor of human BSEP (Ki=2.4 μM; α=2.4), rat Bsep (Ki=5.6 μM; α=34), and human MRP4 (Ki=12.9 μM; α=2.1), and a weak inhibitor of human NTCP (IC50=126.5 μM) and rat Ntcp (IC50=48.4 μM). Employing only AMG 009-specific data, modeling of 100 mg AMG 009 bid for 14 days predicted serum ALT elevations >3X upper limit of normal (ULN) in 17% of the human SimPops™; this was similar to the clinically observed incidence of 12.5% (1 out of 8). DILIsym® recapitulated the clinically observed ALT dynamics (e.g., delayed hepatotoxicity presentation, recovery after discontinuation). Predicted incidence of Hy’s Law cases (ALT>3X ULN and bilirubin>2X ULN) was 1.4%, whereas no Hy’s Law cases were observed among 8 individuals administered 100 mg. At 25 and 50 mg dosing, minimal toxicity was predicted (0 and 2% of ALT>3X ULN, respectively; no Hy’s Law cases), consistent with the clinical data. Conversely, no hepatotoxicity was predicted in the rat SimPopsTM administered 1500 mg/kg/day AMG 009 for 1 month, consistent with the preclinical data.

Conclusions: DILIsym® predicted dose-dependent, delayed AMG 009 hepatotoxicity in humans but no hepatotoxicity in rats, consistent with observed clinical and preclinical data. Mechanistic modeling is a useful approach to translate in vitro bile acid transporter inhibition data to clinical hepatotoxicity.

Sixth American Conference on Pharmacometrics (ACOP), Octoer 4-8, 2015, Arlington, Virginia

By Kyunghee Yang, Jeffrey L. Woodhead, Ryan E. Morgan, Paul B. Watkins, Brett A. Howell, Scott Q. Siler