Purpose of review: Idiosyncratic drug reactions pose a significant clinical threat and hamper drug development. The idiosyncratic nature of these reactions has made mechanistic studies exceedingly difficult, and yet without a better understanding of the mechanisms involved it is unlikely that much progress can be made in dealing with the problem. Several working hypotheses have been used to study these reactions, but none fits all of the characteristics that are observed. Borrowed from immunology, the danger hypothesis has most recently been used to explain several characteristics of these reactions. The present review describes the danger hypothesis and compares it with previous hypotheses to determine how well each fits with the observed characteristics of the reactions.
Recent findings: Slow progress in the field continues and it is important to use new observations, such as identifying T cells that recognize drugs in the absence of reactive metabolite formation, to test and refine the working hypotheses. However, the development of animal models of idiosyncratic drug reactions as well as progress in basic immunology and genomics are likely to accelerate progress in this area in the near future.
Summary: No one model fits the characteristics of all idiosyncratic drug reactions; however, the danger model provides a new perspective and suggests avenues of research that have the potential to increase our ability to predict and prevent such reactions significantly.