A Mini Review of Mammalian Toxicity (Q)SAR Models

Abstract

This paper is based on an in-depth review performed by the European Chemicals Bureau of the European Commissions Joint Research Centre in support of the development of technical guidance for the implementation of the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) legislation, and is one of a series of minireviews in this journal. Compared with the QSAR modelling of ecotoxicity, the modelling of mammalian toxicity is more complicated. This is reflected by the relatively small number of QSAR studies on mammalian toxicity published in the literature. The main problem is related to wide variations in the accuracy of the in vivo data, in the organisms used, and also in the complexity and poor understanding of the mechanisms involved, especially for long-term effects. Nevertheless, the efforts already made to develop QSAR models for mammalian toxicity demonstrate the usefulness of the approach not only for predictive purposes but also for a better understanding of the multiple mechanisms involved in the toxicity. However, there is a need to better characterise existing models in accordance with the OECD QSAR validation principles, to develop new models, and to investigate the applicabilities of individual models in the regulatory context, such as for the purposes of the REACH legislation.

Are the Chemical Structures in Your QSAR Correct?

“…Thus, there has been a great deal of research undertaken in recent years to develop techniques to predict toxicological values without the use of animals. Some of these techniques are semi-empirical, such as (quantitative) structure -activity relationships (Q)SARs [3,4], decision trees [5 -7], and techniques described by Dudek et al [8]. There are other techniques that are based on biological processes, such as those developed to model the dose -response relationship including bioinformatics, chemoinformatics, and molecular modeling [9].…”

Prediction of Acute Rodent Toxicity on the Basis of Chemical Structure and Physicochemical Similarity

“…As a result of the complex and as yet incomplete understanding of the mechanisms underlying acute systemic toxicity, the number of publications devoted to the QSAR modelling of this endpoint is rather small. [2] Moreover a recent detailed analysis of 150 QSAR models led to the conclusion that many available models have only limited usefulness for the purpose of LD 50 predictions due to their poor or modest statistical quality or because they were obtained from limited data sets. [3] Most of the studies in the literature correlate acute mammalian toxicity values with hydrophobicity [4,5] although it is widely recognized that a description of this complex endpoint by only transport-related parameters alone is insufficient.…”

Acute intravenous toxicity to mice calculations on the basis local regression models in superoverlapping clusters (LRMSC)

“…Такие модели обсуждаются в обзорах [см. обзоры 14,15]. При этом было выявлено, что подавляющее большинство из 150 опубликованных КССА моделей острой токсичности по отношению к теплокровным имеет весьма ограниченную пользу, что связано с довольно скромным статистическим качеством этих моделей и ограниченным числом соединений в рассмотренных выборках.…”