One of the major reasons drugs fail in clinical trials is unexpected toxicity due to metabolism that was not accurately predicted in animal model systems or current cell culture models. Extensive efforts are used during lead optimization to avoid liver toxicity in particular and to accurately predict toxic bioactivation products that may be produced over time during the course of drug treatment. Currently the methods to determine whether a parent molecule produces a toxic (or bioactive) metabolite involves a combination of in silico modeling combined with technically challenging in vitro trapping or LC-MS studies of radiolabeled parent drugs followed by large scale HPLC purification and testing. In other words, a significant effort to both identify the metabolites isolating them is a prerequisite to determining their effects.
A highly predictive cell-based assay (e.g. SciFlow technology) could be used to categorize leads for bioactive toxic metabolites without a priori knowledge of the chemical structures. Successful implementation would therefore provide significant cost savings compared to current methods.