Assessing the Use of Quantitative Structure-Activity Relationship Models in Previously Unevaluated Analysis of Alternatives in the European Union Registration, Evaluation, Authorization and Restriction of Chemicals

Untested chemicals released into the market could have harmful effects on human health and the environment. Non-testing methods such as quantitative structure-activity relationship (QSAR) models may prevent these harmful consequences. However, without a meaningful evaluation of QSAR usage and proper documentation under the European Union’s (EU) Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) authorization process, the European Chemicals Agency (ECHA) will continue to make decisions as to whether to authorize Annex XIV chemicals are based on uncertain quality of these QSAR predictions. The four major research questions of this study are: 1) To what extent are QSARs used in AoAs to support, eliminate or evaluate in a weight of evidence (WoE) context an alternative to an Annex XIV chemical?; 2) How did applicants document their QSAR use in AoAs?; 3) How was WoE using QSAR predictions in AoAs used to assess priority endpoints?; 4) How can battery ITS QSAR models further the evaluation of potentially harmful chemicals in AoAs? In order to conduct an analysis on QSARs, it was important to first have a firm understanding of how regulatory models work in the European Union. To become familiar with this process, two regulatory models were built at the Technical University of Denmark (DTU) for predicting agonism and antagonism of the Constitutive Androstane Receptor (CAR) for future upload to the Danish (Q)SAR Database. Applying this knowledge, data were collected from 189 AoAs through May 2017 to assess QSAR usage in AoAs, however, low numbers suggested that QSARs may not have been fully utilized. To explore possible reasons behind these statistics, an assessment of proper documentation of QSAR predictions in AoAs well as a review on the completeness of WoE using QSARs for higher-tier endpoints were performed. Results indicated that several completeness criteria were not met, including one of our priority criteria, structural analogues. In addition, only a limited number of AoAs used Woe with QSARs. A comparison of WoE using QSARs from the AoA sample with Danish EPA battery ITS QSAR predictions suggested that current use of single QSAR models continues to be limited