Development of an in vitro test battery for a test system to predict human drug-induced liver injury

Abstract

Drug-induced liver injury (DILI) is a major concern due to its poor predictability. Recently, we have developed an in vitro/in silico test system for the prediction of human DILI in relation to oral doses and blood concentrations. Additionally, two indices, the toxicity separation index (TSI) and the toxicity estimation index (TEI) were introduced for the quantitative evaluation of a test system and its input parameters. In this PhD-thesis, I studied whether extending the in vitro test battery of the test system, so far consisting of a cytotoxicity test in primary human hepatocytes, by additional functional readouts would lead to improved performance and thus allow a more accurate prediction. In total, three different approaches that address putative DILI-relevant mechanisms were explored. In the first approach, the influence of a bile acid mix on the cytotoxicity of in total 18 compounds in cultivated primary human hepatocytes was investigated. In summary, increased and decreased susceptibility to both hepatotoxic and non-hepatotoxic substances was observed with the addition of bile acids, which did not improve the TSI (0.79 -> 0.77) nor the TEI (0.73 -> 0.69) compared to cytotoxicity without additional bile acids. In the second approach, an assay was developed and evaluated that measures the inhibition of bile acid export carriers in primary human hepatocytes. In total 36 compounds were tested with the transport inhibtion assay and the cytotoxicity assay. In conclusion, the assay is able to detect bile acid export carrier inhibition and integration into the in vitro test battery improved TSI (0.77 -> 0.89) and TEI (0.69 -> 0.83) compared to cytotoxicity. In a third approach, intracellular lipid accumulation in HepG2 cells was investigated for a total of 60 compounds. Addition of lipid droplet accumulation to cytotoxicity improved TSI (0.74 -> 0.80) and TEI (0.67 -> 0.81). In summary, three assays were developed for the in vitro test battery of a test system to predict drug-induced liver injury. Quantitative analysis revealed that two of the three assays lead to improved separation of hepatotoxic and non-hepatotoxic compounds, as well as improved estimation of in vivo relevant blood concentrations. These improvements allow more accurate prediction of DILI by the test system.