LC-MS untargeted metabolomic analysis of drug-induced hepatotoxicity in HepG2 cells

Hepatotoxicity is the number one cause for agencies not approving and withdrawing drugs for the market. Drug‐induced human hepatotoxicity frequently goes undetected in preclinical safety evaluations using animal models. Human‐derived in vitro models represent a common alternative to in vivo tests to detect toxic effects during preclinical testing. Most current in vitro toxicity assays rely on the measurement of nonspecific or low sensitive endpoints, which result in poor concordance with human liver toxicity. Therefore, making more accurate predictions of the potential hepatotoxicity of new drugs remains a challenge. Metabolomics, whose aim is to globally assess all the metabolites present in a biological sample, may represent an alternative in the search for sensitive sublethal markers of drug‐induced hepatotoxicity. To this end, a comprehensive LC‐MS‐based untargeted metabolite profiling analysis of HepG2 cells, exposed to a set of well‐described model hepatotoxins and innocuous compounds, was performed. It allowed to determine meaningful metabolic changes triggered by a toxic insult and gave a first estimation of the main toxicity‐related pathways. Based on these metabolic patterns, a partial least squares‐discriminant analysis model, able to discriminate between nontoxic and hepatotoxic compounds, was constructed. The approach described herein may provide an alternative for animal testing in preclinical stages of drug development and a controlled experimental approach to gain a better understanding of the underlying causes of hepatotoxicity.