薬剤性肝障害のリスク評価におけるin vitroモデルの有効性の検証

抄録

Drug-induced liver injury (DILI) is one of highest causes of drug attrition. The plasma Cmax total at efficacious doses is reported to be a major driver for DILI predictions (Shah et al, Toxicol Sci 2015). Recently, scoring system combining physicochemical properties and Cmax-corrected IC50 of in vitro assay endpoints was proposed for DILI prediction (Aleo et al, Chem Res Toxicol 2019). Because it is challenging to estimate human plasma Cmax in early phase of drug development, we assessed the use of multiple in vitro assay endpoints without Cmax-correction and compare predictive power with published physicochemical property-based DILI risk classification (BDDCS; Benet et al, AAPSJ 2011) to propose potential workflow to prioritize drug candidates in early drug discovery phase.

We assessed the utility of HepG2- and human hepatocyte spheroids-based cytotoxicity assay, HepaRG-based GSH depletion assay and liver humanized mice-derived human hepatocytes-based bile acid transport inhibition assay. Eighty licensed drugs were evaluated. The results showed in vitro assays in combination were able to predict DILI with sensitivity range of 54-61%; specificity range of 77-84%. Most of the false positive drugs showed the Cmax values < 1μM; sufficiently high margins. In vitro assays and BDDCS detected different DILI drugs as positive with overwrap. Our data indicated in vitro assays and/or BDDCS could contribute to drug candidate prioritization in early drug development, and correction of assay results with human exposure might be effective to improve predictive power when available.