? DESCRIPTION (provided by applicant): The high failure rate of drugs in late stage development, nearly 90%, is a symptom of the inadequacy of pre-clinical animal models to accurately predict human biology. One of the major reasons drugs fail in the clinic is inaccurate prediction of the therapeutic index: which is the dose range where drugs are effective and also not toxic. Failure to identify metabolism produced toxicities in preclinical efforts lead to unwelcome and highly expensive surprises in clinical trials and even post-marketing of new drugs. SciKon has created a serial capillary system in a standard cell culture format that enables researchers to apply drugs to cells in culture using a more in-vivo-like exposure dynamic. In turn a compartmentalized dynamic gradient enables query of each internal serial system for evaluating cellular mechanisms in both a concentration and time-resolved manner. This proposal describes using this system to query cytotoxicity of drugs to primary human hepatocytes as a function of the real time changes in parent: metabolite ratio. Our preliminary tests suggests the feasibility of creating a screening assay that distinguishes between whether a test chemical is metabolized to something toxic or not without a priori understanding of a drugs' overall metabolism or metabolite ID. The aims of this proposal are to 1) quantitate metabolites of common test compounds as a function of both time, parent concentration, and location within the system and 2) to evaluate a 11 compound training set of drugs with known hepatotoxicity in the clinic as well as the metabolic component of that toxicity. Such an assay can be inserted into the drug development pipeline much earlier than current hepatocyte metabolism studies where knowing the metabolite ID is a prerequisite, thus reducing the overall cost of drug development and increasing the chances of success in the clinic. Furthermore, this assay would also apply to chemicals for industrial, consumer product, cosmetics, and foods.