The definition of hormesis and its implications for in vitro to in vivo extrapolation and risk assessment.

This article comments on some of the basic questions put forward in state-of-the-art discussions on hormesis. There seems to be a need for a better definition of the concept itself and reconsideration of whether all biphasic dose-response curves should be considered representative for hormesis. Hormesis may be restricted to phenomena that proceed by mechanisms that are broadly generalizable and represent possibly beneficial overcompensation in response to an adverse stimulus. Using the concept that hormesis is defined as such, the biphasic effect of quercetin on cell proliferation, but also several other receptor-mediated biphasic dose-response phenomena, should not be related to hormesis. Taking into account hormesis in the procedures for risk assessment on compounds characterised by a threshold for the adverse effect is another matter for considerable debate. In our opinion, this would require the reduction of safety factors, providing the possibility for beneficial hormesis-type effects for some people, at the cost of increased chances on adverse effects for other parts of the population. Whether this is a proper way forward remains to be discussed. Improvement of risk assessment strategies may include taking into account biphasic dose-response curves, but should rather start with the consideration of proper physiologically based pharmacokinetic (PBPK) models for better extrapolation of differences in toxicokinetics going from high- to low-dose exposure, as well as taking into account kinetics for gene repair systems. Without considering in vivo toxicokinetics in the in vitro models, extrapolation from in vitro biphasic dose-response curves on cell proliferation to in vivo cell proliferation is difficult to do. Altogether, it is concluded that hormesis is an important phenomenon, especially from the scientific point of view, but that its consequences for risk assessment and the possibilities for in vitro to in vivo extrapolation may remain limited without additional mechanistic insight.