SC3.4 Integrating DILI Hazards in Predicting Toxicity of Drug Candidates: Assessment and Management of Human Risk in the Pharmaceutical Industry

John Gerald Kenna , Safer Medicines Trust, Macclesfield, United Kingdom
Many drugs cause infrequent and idiosyncratic drug-induced liver injury (DILI) in humans, which is not predicted by animal safety studies. Human idiosyncratic DILI is a leading cause of attrition late in clinical development, failed registration, withdrawal of licensed drugs and drug-induced serious human ill-health. In addition, numerous candidate drugs cause reproducible and dose-dependent toxicity to the liver which is observed during their nonclinical safety evaluation in animals and leads to compound attrition. Dose-dependent liver toxicity in animals and idiosyncratic human DILI pose major problems to drug developers, drug prescribers and drug-treated patients. Therefore identification of compounds which have high propensity to cause these toxicities is necessary, and should be undertaken prior to animal safety studies and commencement of clinical trials. This can be achieved by use of in vitro assays which evaluate mechanisms by which drugs can initiate liver toxicity. An especially important mechanism is formation of chemically reactive metabolites. Absolute elimination of reactive metabolite formation is impractical, and may be undesirable – for example, when covalent binding to a target enzyme is required for efficacy. Hence it is important to determine when bioactivation raises concern and must be reduced, or avoided, and when it does not. An especially useful approach is quantification of covalent binding of radiolabelled drug to human hepatocyte proteins. When interpreting covalent binding data it is important to take account of fractional in vitro metabolic turnover, plus the daily dose of the drug in patients. Drugs with high “covalent binding body burdens” have high human DILI concern (see: Chem Res Toxicol 2012; 25:1616). Liver injury can also be caused by initiating mechanisms other than metabolic bioactivation. These include mitochondrial injury, metabolism-independent cell cytotoxicity and inhibition of the hepatic Bile Salt Efflux Pump (BSEP), which also need to be considered when assessing drug safety. Ideally, a single in vitro assay platform would be used to evaluate these additional mechanisms alongside metabolic bioactivation. In practice, multiple assays which are needed, which each assess different mechanisms. The need to interpret and integrate data provided by multiple assays poses a major challenge. Effective interpretation of in vitro toxicity assay data requires account to be taken of assay potency (typically expressed as EC50 or IC50 values) and also human drug exposure in vivo. Ideally, the concentration of drug within liver cells should be used in the data analysis. However, usually this is not known and so estimated or experimentally determined drug concentrations in plasma must be used instead. A convenient approach is to calculate the ratio between the total (protein bound and unbound) plasma drug concentration and the in vitro assay IC50 (or EC50) value, and to use a ratio of 0.1 as the “threshold of concern”. A hazard matrix approach can then be used to integrate data provided by multiple assays, and thereby to identify drugs which have high human DILI propensity with excellent sensitivity and specificity (see: J Pharmacol Exp Ther 2015;352:281).