Improved non-animal testing for chemical safety analysis
REACH aimed to ensure the safe and responsible production and use of chemicals to protect human health and the environment. The work was carried out in compliance with EU regulation, Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). Members of the CADASTER project employed optimised intelligent/integrated testing strategies (ITSs) by combining in silico modelling and in vitro testing methods. Lucid application criteria and guidelines were also provided to facilitate their adoption by industries and small and medium-sized enterprises (SMEs). Several significant milestones were achieved within the scope of the project. Four classes of chemicals — flame retardants, fragrances, perfluorinated chemicals (PFCs), and triazoles(TAZs) and benzotriazoles (BTAZs) — were selected for case study. A harmonised user-friendly database was developed that encompassed comprehensive information on these chemicals and representative models. For accurate risk assessment and chemical prioritisation for experimental testing, properties evaluated included chemical toxicity, bioaccumulation and biodegradability. The team created a priority list of brominated flame retardants, PFCs and (B)TAZs using similarity analysis and multivariate ranking methods. Study parameters included several toxicological endpoints, endocrine disruption (ED) potency, melting point (MP), boiling point (BP) and environmental behaviour endpoints. Some of the novel non-animal testing options developed were quantitative structure activity relationships (QSARs), read-across (interpolating information on related compounds), category approaches and exposure-based waiving (proof that chemical exposure concentrations are below the no-observed effect level). These predictive ad hoc models were validated for accuracy, reproducibility, efficacy and applicability by comparing predicted and actually measured data values of the chemicals. Individual QSAR and quantitative property relationship models were combined to develop consensus models for MPs, BPs, fragrances, PFCs and BTAZs using same endpoint or chemical class as criteria. Researchers provided clear application guidelines for ITSs by characterising uncertainty, variability and model sensitivity methodologies. The (Q)SAR Model Reporting Format (QMRF) enabled the internal and external validation of models with a user-friendly graphical representation to improve understanding. Remote access of the web services was also enabled within standalone decision support systems (DSSs) for external users. besides training via workshops to relevant stakeholders, comprehensive chemical data and models are freely provided via the project website and relevant database websites. The training provided also covered the development of new models for other chemical compound groups with limited or unreliable data. Research results were disseminated through newsletters, presentations, conferences, workshops, the project website and scientific publications.