Abstract

Troglitazone (TGZ) caused life-threatening drug-induced liver injury (DILI) in diabetic patients, whereas the next in class, pioglitazone (PGZ), has rarely been associated with DILI. Inhibition of bile acid transport, which may result in accumulation of toxic bile acids in hepatocytes, is one proposed mechanism of TGZ-mediated hepatotoxicity. However, PGZ is a more potent inhibitor of the bile salt export protein (BSEP) than TGZ based on in vitro membrane vesicle transport studies. In the current study, the hepatotoxic potential of TGZ and PGZ due to interference with bile acid homeostasis was investigated using DILIsym®, a systems pharmacology model of DILI. Experimentally measured inhibition constants of TGZ, TGZ-sulfate, and PGZ for multiple bile acid transport proteins were employed to simulate the altered bile acid disposition and subsequent DILI in humans. In the virtual human population (Simpops™), administration of 200–600mg/day TGZ for 6 months resulted in delayed increases in serum alanine transaminase (ALT) > 3X upper limit of normal (ULN) in 0.3–5.1% of the population, with concomitant elevations in serum bilirubin > 2X ULN in 0.9–3.6% of the population. The simulated time to peak ALT was 116±60 days. These results were similar to observations from the clinical trials where 200–600mg/day TGZ elicited serum ALT elevations > 3X ULN in 1.9% of treated patients with time to peak ALT of 147±86 days. No hepatotoxicity was predicted in the SimpopsTM after administration of clinically relevant doses of PGZ (15-45mg/day) for 6 months, consistent with the clinical observations. In summary, mechanistic modeling based only on bile acid homeostasis adequately predicted the incidence and delayed presentation of TGZ hepatotoxicity, and correctly predicted relative liver safety of PGZ. These results demonstrate the utility of systems pharmacology models that integrate physiology and experimental data to evaluate DILI mechanisms and identify potential risk factors for DILI. Importantly, these mechanistic models may be useful to prospectively predict the hepatotoxic potential of new drug candidates.

19th North American ISSX & 29th JSSX Meeting (Dedicated in Honor of Yuichi Sugiyama, Ph.D.), October 19-23, 2014, San Francisco, CA

By Kyunghee Yang, Jeffrey L Woodhead, Paul B Watkins, Brett A Howell, and Kim LR Brouwer