Characterisation of the REACH Pre-Registered Substances List by Chemical Structure and Physicochemical Properties

Abstract

In the European Union, the registrants of chemical substances under the REACH legislation are explicitly encouraged, and even required, to use non-testing methods as a means of identifying the presence or absence of hazardous properties of substances in order to meet the information requirements of REACH while at the same time minimising testing on vertebrate animals. The need to use non-testing methods or other alternative (nonanimal) methods such as in vitro tests, has led to the development and implementation of Integrated Testing Strategies based as far as possible on the integrated use of non-animal data. The use of non-testing methods within such strategies implies the need for computational tools and a structured workflow to facilitate their application. The list of pre-registered substances (PRS) published by the European Chemicals Agency includes chemicals that industry may to register in accordance with the deadlines specified in the REACH legislation. The PRS list does not include information on chemical structures, which are a prerequisite for the development and application of non-testing methods. Therefore, in order to facilitate the implementation of non-testing methods for the regulatory assessment of REACH chemicals, the Computational Toxicology Group within the Joint Research Centre (JRC) has: • generated structures for the PRS were by using the ACDLabs Name-to-Structure (NTS) software and validated them with a random sample • processed the structures to generate substance identifiers such as IUPAC name, InChI codes and SMILES strings • processed the structures to obtain information on chemical characteristics suitable for a preliminary assessment of the hazard and exposure • indicated the availability of experimental toxicological data with DSSTOX and FOOTPRINT tags • created a “QSAR-ready” data file to support the application of non-testing methods, such as QSARs The application of ACD NTS resulted in a high rate of yield for the generation of structures (85%) for monoconstituent substances with a high reliability – in total, about 80,000 structures were generated. By comparing these results with inventories of structures available from other publicly available sources of information, additional high quality structures including precise information on stereochemistry were generated. A quality review resulted in the assignment of quality labels to the structures and in the further checking of about 5500 structures. To support QSAR predictions and to estimate key physicochemical properties of the substances, these structures were processed to obtain an inventory of PRS parent substances, which serves as a standardised input for computational tools containing about 62,000 records. For these parent compounds the key features of the structures were calculated and key physicochemical properties were estimated using EPISUITE, Pipeline Pilot and ADMET Predictor. This chemical characterisation of the parent substances can be used to support a preliminary assessment of hazard and exposure. The data highlight the importance of ionisation to predict hazard and exposure for about 40% of the substances in the inventory.