Toxicological methods, developing better

The major FDA concern came to be better comprehension of diethylene glycol s toxicology. The imminent trial in court required this. In a more basic sense, the crisis made FDA scientists aware of inadequacies in the state of the discipline. In constant contact with their peers at the AMA and at the University of Chicago and Johns Hopkins, a team of FDA scientists launched a project that "developed the first valid process for determining the comparative toxicity of compounds, a statistically based and legally defensible process that opened the door to modern toxicological testing methods" (77). 

There are many circumstances in which the only information we can develop on toxic hazards and dose-response relationships derives from experiments on laboratory animals. The example of the food additive, presented in the opening pages, is just one of many circumstances in which condition A involves animal toxicology data, and condition B involves a human population, almost always exposed at small fractions of the dose used in animals, and sometimes exposed for much larger fractions of their lifetime than the animals, and even by different routes. Extrapolations under these circumstances should cause individuals trained in the rigors of the scientific method to seek some form of psychological counsel, or, better yet, to return to the laboratory. 

Quantitative structure-activity relationship (QSAR) dates back to the nineteenth century and is a computer-based tool that attempts to correlate variations in structural or molecular properties of compounds with their biological activities. These physicochemical descriptors, which include parameters to account for hydrophobicity, topology, electronic properties, and steric effects, are determined empirically or, more recently, by computational methods. The premise is that the structure of a chemical determines the physiochemical properties and reactivities that underlie its biological and toxicological properties. Being able to predict potential adverse effects not only aids in the designed development of new chemicals but also reduces the need for animal testing. It may ultimately or potentially lead to better... 

There has been some research into combining the output from several computer systems that make predictions for the same toxicological endpoints on the basis of chemical structure and physicochemical properties. Until fairly recently researchers favoring numerical or nonnumerical methods have tended to work separately and to develop schemes better suited to one or other of the approaches. 

---