Assessing risks and mechanisms of idiosyncratic drug toxicity by fingerprints of cell signaling responses.

Idiosyncratic drug-induced liver injury (DILI) is the leading cause of post-marketing drug withdrawal. Here, we describe a straightforward DILI liability assessment approach based on fingerprinting cell signaling responses. The readout is the activity of transcription factors (TF) that link signaling pathways to genes. Using a multiplex reporter assay for 45 TFs in hepatocytic cells, we assessed TF activity profiles (TFAP) for 13 pharmacological classes. The TFAP signatures were consistent with primary drug activity but transformed into different, off-target signatures at certain concentrations (COFF). We show that the off-target signatures pertained to DILI-relevant mechanisms, including mitochondria malfunction, proteotoxicity, and lipid peroxidation. Based on reported plasma concentrations in humans (CMAX), drugs do not reach the off-target thresholds in vivo, consistent with the lack of overt toxicity in the population. However, DILI liability drugs were dangerously close to the off-target thresholds. We characterized this closeness by the COFF/CMAX ratio termed the safety margin (SM). Most-DILI-concern drugs invariably showed smaller safety margins than their less-concern counterparts in each pharmacological class and across classes (median SM values of 6.4 and 212.7, respectively (P<0.00015)). Therefore, the TFAP approach helps to explain idiosyncratic drug toxicity and provides clear quantitative metrics for its probability and the underlying mechanisms.