Title: An investigation into the use of computational and in vitro methods for acute systemic toxicity prediction
Authors: NORLEN HedvigBERGGREN KatarinaWHELAN MauriceWORTH Andrew
Publisher: Publications Office of the European Union
Publication Year: 2012
JRC N°: JRC72736
ISBN: 978-92-79-26058-2 (print)
978-92-79-21708-1 (pdf)
ISSN: 1018-5593 (print)
1831-9424 (online)
Other Identifiers: EUR 25473 EN
OPOCE LB-NA-25473-EN-C (print); LB-NA-25473-EN-N (online)
URI: http://publications.jrc.ec.europa.eu/repository/handle/JRC72736
DOI: 10.2788/44144
Type: EUR - Scientific and Technical Research Reports
Abstract: We have assessed the abilities of five alternative (non-animal) approaches to predict acute oral toxicity, a toxicological endpoint relevant to multiple pieces of legislation on chemicals and consumer products. In particular, we have investigated four QSAR models (ToxSuite, TOPKAT, TEST and ADMET Predictor) and one in vitro method (3T3 NRU). Based on a test set of in vitro and in vivo data for 180 compounds, we have characterized the predictive performance of each method when used alone (both for LD50 prediction and acute toxicity classification into three categories), as well as multiple test combinations (batteries) and stepwise testing strategies (for acute toxicity classification into three categories). When used individually, the alternative methods showed an ability to predict LD50 with correlation coefficients in the range from 49% to 84%, and to classify into three toxicity groups with accuracies in the range from 41% to 72%. When the alternative methods were combined into batteries or testing strategies, the overall accuracy of prediction could reach 76%. We also illustrate how different combinations of methods can be used to optimize sensitivity or specificity.
JRC Directorate:Institute for Health and Consumer Protection Historical Collection

Files in This Item:
File Description SizeFormat 
lbna25473enn.pdf1.14 MBAdobe PDFView/Open

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.