Senior Research Fellow, Pfizer Inc., USA
Toxicology has decided that to become a modern discipline, it must focus more on molecular biology and molecular toxicology. The US Government’s Toxicology in the 21st Century program is the epitome of this approach, focusing on molecular changes in model cell systems as representative of the critical response in whole animals or populations. This is toxicology’s equivalent of high-throughput drug discovery, having a reductionist focus on representative target gene or protein changes in model systems. However, there is a growing awareness in the drug discovery community that this approach is less successful than using a phenotype-based approach. Indeed, few or no drugs have been found and approved using the high-throughput, rational-design approach. The emerging appreciation that many gene-disease associations were really ‘noise discovery’ (Ioannidis, 2005) provides additional caution about a gene-focused approach to understanding mechanisms of toxicity.
An additional layer of complexity comes from recognising that environmental compounds (and many medicines, too) do not act as a single place in the cell. They never have just one target, and with higher doses they hit more targets. Phenotype aggregates all those differing molecular changes into one set of histologic presentations.
If we believe that looking at phenotype will better describe what sorts of changes are adverse, and which is instead, biological noise, then the way forward becomes clear: we must create a database of exposures of whole animals to epigenetic-modifying compounds and examine the galaxy of results. If a new pattern emerges which is distinct from that pattern created by other reproductive toxicants, then that provides a focus for future work, and we can consider whether a refined focus will save time or money. It may be that all epigenetic modifiers, regardless of specific mechanism, will produce effects similar to other reproductive toxicants, so that simply by knowing their mechanism allows a prediction of their in vivo activity. Science is never that simple, though, so we can suspect it is much more likely that some epigenetic compounds will be reproductive toxicants and others will be without reproductive toxicity except at significant multiples of exposure (which will thus likely be the result of undetermined off-target effects). In this case, simply knowing the mechanism will not be enough to allow us to predict their toxicity.
Only by accumulating a database correlating known (or intended) mechanism with in vivo effects in a model species or two will allow us to test for any association between mechanism and outcome. Critically, this will need to contain enough different exposures to allow the conclusion to be robust……or at least not to be misleading.