Approaches to the study of drug-induced liver injury.

Retrospective studies of administrative databases have led to highly variable estimates of the incidence and natural history of drug-induced liver injury (DILI) and the agents implicated. In contrast, prospective multicenter registry studies allow for more accurate phenotyping of individual patients with DILI, as well as for collection of biological samples for mechanistic studies. In addition to improved diagnostic and prognostic biomarkers, standardized causality assessment tools are needed, as well as population-based studies that represent the full spectrum of disease severity. This would facilitate further research into the pathogenesis of DILI, leading ultimately to the prevention of this condition.