In order to know how animal experiments can be replaced or reduced, we need to know how results from methods making less or no use of animals compare with results from animal experiments, as predictors of health effects in humans. Such comparisons should be made for the purpose of risk management. This means that fairly rough predictions of the magnitude of risk are sufficient and (according to the precautionary principle) that false positives are more tolerable than false negatives. Several studies have been made of the correlation between different types of toxicity and different test methods. However, in spite of these individual studies no effort seems to have been made to develop the study of toxicological correlations as a general area of knowledge. Therefore, this is still largely an unexplored field lacking a comprehensive approach.
We will develop optimised testing strategies, making use of available information about correlations between results in different test systems and, in particular, of new information about correlations between test outcomes that can be obtained through correlation studies on literature data. This project aims to propose a test strategy that is efficient and simplified, and that does not make use of experimental animals, and hence is suitable to perform, as a first step, on a larger number of substances for which data are lacking.