WEDNESDAY, 16 APRIL 2014Whatever your opinions about its morality, the stringent animal testing required in drug development ensures that human side effects are almost always identified by previous animal trials. When the fialuridine trial ended so disastrously, scientists were therefore all the more shocked because rigorous experiments in mice, rats, dogs and monkeys had failed to give any forewarning of the acute liver failure that would occur.
Prior to the publication of this work, we had no reassurance that we would be able to detect similar toxicity of future drugs from animal models. But Peltz and colleagues found a mouse model which can predict this side effect of fialuridine, and possibly those of other liver drugs too. The researchers used a mouse which has a humanised liver - a liver in which 90% of mouse cells are replaced by human ones. They tested this model by administering each mouse either with fialuridine or with a control drug which did not cause human liver toxicity, and found that the mice experienced liver failure in the case of fialuridine only. Given that the symptoms were very similar to those seen in humans, and that the response was specific to toxic fialuridine, these results allow optimism that the model could successfully be used to identify human-toxic liver drugs such as fialuridine before they go to human clinical trials.
The disadvantage of using mice with a humanised liver is that they must be immunocompromised to prevent rejection of the organ, meaning that immune-mediated toxicity cannot be detected by the model. However, most drugs are likely to harm the liver directly instead, because of its role as the body’s main detoxifying organ. Developing drugs which do not induce direct toxicity on the liver can be difficult. The improved early detection of such side effects in animal trials is therefore a positive step towards more efficient drug development, as well as making human clinical trials safer and preventing future deaths.
Full paper available at: http://www.plosmedicine.org/article/info%3Adoi%2F10.1371%2Fjournal.pmed.1001628