TAK-994 mechanistic investigation into drug-induced liver injury

Abstract

The frequency of drug-induced liver injury (DILI) in clinical trials remains a challenge for drug developers despite advances in human hepatotoxicity models and improvements in reducing liver-related attrition in preclinical species. TAK-994, an oral orexin receptor 2 agonist, was withdrawn from phase II clinical trials due to the appearance of severe DILI. Here, we investigate the likely mechanism of TAK-994 DILI in hepatic cell culture systems examined cytotoxicity, mitochondrial toxicity, impact on drug transporter proteins, and covalent binding. Hepatic liabilities were absent in rat and nonhuman primate safety studies, however, murine studies initiated during clinical trials revealed hepatic single-cell necrosis following cytochrome P450 induction at clinically relevant doses. Hepatic cell culture experiments uncovered wide margins to known mechanisms of intrinsic DILI, including cytotoxicity (>100× Cmax/IC50), mitochondrial toxicity (>100× Cmax/IC50), and bile salt efflux pump inhibition (>20× Css, avg/IC50). A potential covalent binding liability was uncovered with TAK-994 following hepatic metabolism consistent with idiosyncratic DILI and the delayed-onset clinical toxicity. Although idiosyncratic DILI is challenging to detect preclinically, reductions in total daily dose and covalent binding can reduce the covalent body binding burden and, subsequently, the clinical incidence of idiosyncratic DILI.

Keywords: bile salt efflux pump; covalent binding; drug discovery; drug-induced liver injury; hepatotoxicity.

Fig. 1.

Preclinical compounds showing liver toxicity. Thirty-eight (74.5%) compounds from the preclinical pipeline were identified as having liver toxicity from assays in HepG2 or human liver microtissues (hLiMT), whereas 13 (25.5%) of compounds tested were undetected (shown at the midpoint on the graph). Eighteen (35.3%) compounds were detected in both assays, but only 5 (9.8%) and 15 (29.4%) compounds were detected in separate HepG2 and hLiMT assays. IC₅₀, half-maximal inhibitory concentration.

Fig. 2.

TAK-994 preclinical cytotoxicity studies. (A) TAK-994 did not cause HepG2 cytotoxicity or mitochondrial toxicity. Positive controls were rotenone and staurosporine; n = 2. ATCT72, 72 h in glucose medium; CT24GLU or CT24GAL, 24 h in glucose or galactose medium, respectively. (B) TAK-994 liver microtissue toxicity acts directly on hepatocytes. (C) TAK-994 treatment for 24 h showed no cytotoxicity difference between control HepG2 and CYP3A4 overexpressed cells; n = 3. Unpaired, 2-tailed t-test, *P ≤ 0.05; **P ≤ 0.01 vs vehicle control. IC₅₀, half-maximal inhibitory concentration.

Fig. 3.

TAK-994 preclinical cytotoxicity studies. (A) TAK-994 liver microtissue toxicity not sensitized by intracellular triglycerides. Right, triglycerides measured in free fatty acid (FFA) models (Day 14). Unpaired, 2-tailed t-test, *P ≤ 0.05; **P ≤ 0.01 vs vehicle control. (B) Effect of cholyl-lys-fluorescein (CLF) efflux change after treatment with TAK-994; n = 3; IC₅₀=30.0 μM. (C) TAK-994 inhibits bile salt efflux pump (BSEP)-mediated transport of [3H]TCA. [3H]TCA, [3H(G)]taurocholic acid; IC₅₀, half-maximal inhibitory concentration.

Fig. 4.

TAK-994 liver microtissue toxicity is not sensitized by BSO. The profile of cytotoxic response of (A) TAK-994 and (B) acetaminophen monitored by cellular ATP levels induced by treatment (top) and release of LDH between Days 0 and 5 of treatment (bottom). ATP data shown as mean and SD. LDH data shown as mean and 95% confidence interval, values below 0.5 mU/ml considered as background (bottom dotted line). (C) Cellular GSH is measured in liver microtissues pretreated with BSO and in control microtissues on Days 0, 3, and 14. Unpaired, 2-tailed t-test, *P ≤ 0.05; **P ≤ 0.01 vs vehicle control. Significance between 2 specific samples indicated using a bracket linking the 2 samples. ATP, adenosine triphosphate; BSO, buthionine sulfoximine; GSH, glutathione; LDH, lactate dehydrogenase.

Fig 5.

Cytotoxicity of TAK-994 in the presence and absence of LPS. The profile of cytotoxic response of (A) TAK-994 and (B) trovafloxacin as a positive control was monitored by release of LDH between Days 0 and 5 of treatment (top) and cellular ATP levels on Day 14 of treatment (bottom). LDH data shown as mean and 95% confidence interval, values below 0.5 mU/ml considered as background (bottom dotted line). ATP data shown as mean and SD. Unpaired, 2-tailed t-test, *P ≤ 0.05; **P ≤ 0.01 vs vehicle control. Significance between 2 specific samples is indicated by using a bracket linking the 2 samples. ATP, adenosine triphosphate; LDH, lactate dehydrogenase; LPS, lipopolysaccharide.