Advancing assessments of endocrine disruptors using adverse outcome pathways and novel methodologies

<p dir="ltr">Protecting human and environmental health from adverse effects of endocrine disruptors (EDs) is a high priority in the EU. These substances cause a range of adverse effects, such as reproductive toxicity, neurotoxicity, metabolic disruption, and cancer. If an organism is exposed during a developmental life stage, the effects may also be permanent. Assessing if a substance has endocrine disrupting properties is a lengthy and complex process that requires considerable amounts of animal data. There is a demand to phase-out animal testing in the EU and, simultaneously, there is growing use of alternative methods, like in silico, in chemico, and in vitro methods. Hence, there is a need, and an opportunity, to move toward ED assessments based on mechanistic data from alternative methods. However, there is a lack of methodology to connect these types of novel and complex data to adverse outcomes for use in ED assessments.</p><p dir="ltr">The aim of my doctoral project was to develop and explore methodologies for the assessment of endocrine disruptors based on mechanistic data. The Adverse Outcome Pathway (AOP) framework was crucial to achieve this and was used across all subprojects of the thesis. I explored many complex aspects of the AOP framework, including development and application of AOP network and quantitative AOPs (qAOPs). Literature-based, computational, and experimental approaches were utilized, as well as combinations of these.</p><p dir="ltr">In Study I, I developed a data-driven approach to generate AOP networks. This approach combines a systematic approach to identifying relevant AOPs in the AOP-Wiki with a computational approach to process and filter AOP-Wiki data. The output could then be imported into a software for visualization and analysis of networks.</p><p dir="ltr">In Study II, I conducted and compared a standard ED assessment according to current guidance, to a mechanism-based assessment without in vivo data. A thyroid-related AOP network was utilized for identifying relevant search terms and for mapping collected data to identify a potential Mode of Action. This case study was performed using perfluorooctane sulfonic acid (PFOS) as a model substance, investigating its potential for thyroid disruption and developmental neurotoxicity.</p><p dir="ltr">In Study III, RNA-sequencing of zebrafish embryos exposed to either cadmium (Cd) or 3,3',4,4',5-pentachlorobiphenyl (PCB126) was performed to investigate how transcriptomics data can be coupled to an AOP network. Multiple different data analysis approaches were explored, and their potential connection to an estrogen, androgen, thyroid, and steroidogenesis (EATS) related AOP network was evaluated. Ultimately, I developed and compared a data-driven and an expert-driven approach to connect Gene Ontology (GO) terms from enrichment analysis to key events (KEs) in the network.</p><p dir="ltr">In Study IV, I developed an approach to quantify a Key Event Relationship (KER) based on data from published literature, and applied this approach on a KER linking decreased circulating testosterone levels to decreased sperm count. The approach consists of two parts, (1) A structured search strategy, study reliability assessment, and standardized data extraction to collect and assess data for quantification, and (2) a statistical modelling approach to make the best use of the collected data.</p><p dir="ltr">Through the development and application of AOP networks (Study I and III) I identified several opportunities for improving AOP development and functionality of the AOP-Wiki. The issues identified hinder topological analysis and application of data-driven approaches to connect mechanistic data to the network. In Study II, based on the standard ED assessment, I concluded that PFOS fulfils the scientific criteria as an ED based on thyroid disruption and developmental neurotoxicity. On the other hand, in the mechanism-based assessment, I could only identify endocrine activity and not endocrine-mediated adversity. There was a lack of quantitative understanding of the AOPs in the network, which was essential to inferring adversity based on mechanistic data. Study III revealed that both cadmium and PCB126 potentially have endocrine disrupting properties. Endocrine activity, and adversity likely mediated by EATS or non- EATS modalities, were observed. The data-driven approach to connect transcriptomics data to the AOP network yielded very few results, while the expert-driven approach provided many more connections. Gene expression measurements followed by enrichment analysis may be fit for connecting data to intermediate and late KEs, but not for early KEs and molecular initiating events (MIEs). In Study IV, the quantified KER can be used to predict decreased sperm count based on a decrease in circulating testosterone levels, including some uncertainty in the circulating testosterone levels. The model can potentially be used to reduce or replace animal experiments if coupled to e.g., in vitro testosterone synthesis experiments or physiologically based kinetic (PBK) models.</p><p dir="ltr">To conclude, I have developed and applied novel risk assessment methodologies to maximize the use of mechanistic data for ED assessment. Both opportunities and challenges with these approaches have been explored, and future research needs have been identified. Moreover, endocrine disrupting properties of chemicals not yet classified as EDs (perfluorooctane sulfonic acid , cadmium, and 3,3',4,4',5- pentachlorobiphenyl) were investigated. Altogether, these results pave the way toward non-animal mechanism-based assessment of EDs and improved protection of human health.</p><h3>List of scientific papers</h3><p dir="ltr">I. <b>Wiklund L,</b> Caccia S, Pipal M, Nymark P, Beronius A. Development of a data-driven approach to Adverse Outcome Pathway network generation: a case study on the EATS-modalities. Front Toxicol. 2023 May 9; 5:1183824. <a href="https://doi.org/10.3389/ftox.2023.1183824" rel="noreferrer" target="_blank">https://doi.org/10.3389/ftox.2023.1183824</a></p><p dir="ltr">II. <b>Wiklund L,</b> Pipal M, Weiss J, Beronius A. Exploring a mechanism-based approach for the identification of endocrine disruptors using Adverse Outcome Pathways (AOPs) and New Approach Methodologies (NAMs): A perfluorooctane sulfonic acid case study. Toxicology. 2024 May; 504:153794. <a href="https://doi.org/10.1016/j.tox.2024.153794" rel="noreferrer" target="_blank">https://doi.org/10.1016/j.tox.2024.153794</a></p><p dir="ltr">III. <b>Wiklund L,</b> Wincent E, Beronius A. Using transcriptomics data and Adverse Outcome Pathway networks to explore endocrine disrupting properties of Cadmium and PCB-126. Environ Int. 2025 Mar; 197:109352. <a href="https://doi.org/10.1016/j.envint.2025.109352" rel="noreferrer" target="_blank">https://doi.org/10.1016/j.envint.2025.109352</a></p><p dir="ltr">IV. <b>Wiklund L,</b> Ristovska M, Moe J, Beronius, A. Quantification of the Key Event Relationship "Decreased circulating testosterone leading to impaired spermatogenesis" based on data from peer-reviewed studies. [Manuscript]<br></p>