Notice that the dose-response "curve" for a carcinogenic compound is assumed to be linear and have no lower threshold dose (i.e., any dose is assumed to have a potential to cause cancer).  By contrast, these assumptions are not necessarily made for noncarcinogens.