Towards the Development of Innovative Quantitative Structure-Activity Relationship Models for Human and Ecological Risk Assessment of Chemicals and their Mixtures

Abstract

Humans, animals and the environment are exposed to thousands of chemicals of anthropogenic and natural origin, which may cause adverse health effects. The risks posed by man-made chemicals such as biocides, pharmaceuticals, veterinary products and plant protection products (PPPs) are assessed via Human Health Risk Assessment (HHRA), Animal Health Risk Assessment (AHRA), and Ecological Risk Assessment (ERA). Traditionally, scientific advisory bodies such as the European Food Safety Authority (EFSA), the United States Environmental Protection Agency (US EPA) evaluate the results of toxicity tests on laboratory animals (i.e. in vivo) to derive safe levels of exposure to chemicals. These toxicity tests are considered black-box approaches to RA, as the molecular mechanisms by which chemicals cause adverse health effects and the extent to which effects and dose levels are relevant to other species and individuals remains often, to date, poorly characterised or unknown. Concerns on the relevance and ethics of in vivo tests were raised such as the number of animals, cost and time required for testing: only in 2017, 2.18 million animals were used in EU laboratories to meet legislative requirements in order to ensure chemical safety for human health and/or the environment. Since 2007, the so-called Tox21 strategy in the United States has provided a basis to shift toxicological assessments away from traditional animal tests to in vitro and in silico studies providing a mechanistic basis for chemical toxicity often considered as new approach methodologies (NAMs). In the EU, the Chemical Strategy for Sustainability 2020 aimed to boost innovation for the safe and sustainable use of chemicals. It specifically calls for multidisciplinary research to move away from animal testing. In this thesis, the concept of NAMs, particularly quantitative structure-activity relationship (QSAR) models, has been widely illustrated and discussed through the analysis of five different case studies applied to single chemicals RA (chapters 2 and 4) and mixture risk assessment (MRA) (chapter 3, 5 and 6), respectively, in line with the recent EFSA Guidance Document “MIXTOX” on harmonised methodologies for risk assessment of combined exposure to multiple chemicals. This thesis demonstrated how NAMs such as QSARs can provide scientific advisory bodies and industry with robust animal-free methods to perform future ERA, AHRA and HHRA of regulated chemicals, emerging contaminants and their mixtures. In line with the Tox21 strategy, this work presents practical examples of QSAR model applications in order to i) shift RA to give us a more mechanistic understanding of toxicity and ii) innovate current chemicals safety testing by improving the quality, efficiency and speed of ERA, AHRA and HHRA of single and multiple chemicals. In particular, this thesis provides risk assessors and scientists with sound methodologies and smart strategies such as in silico tools, and responds to the recent call of the EU Chemical Strategy for Sustainability (EC, 2020) aimed to replace animal testing through multidisciplinary research, methods and models, and data analysis capacities.