In Response: Quantitative adverse outcome pathways for prediction of adverse effects—An academic perspective
Stefan Scholz · Environmental Toxicology and Chemistry · 2015
The formulation of the adverse outcome pathway (AOP) concept by Ankley et al. of the US Environmental Protection Agency in 2010 1 had an astonishing impact on the academic (eco)toxicology community. This high interest is certainly related to the link of the AOP concept to a mechanistic understanding of chemical hazards. Elements of this mechanistic understanding had been included in previous concepts, such as biomarkers 2, effect propagation 3, and toxicity pathways 4, and had been discussed for potential use in qualitative and quantitative structure–activity relationship models 5. All these previous approaches included at least partial elements of the AOP concept. However, none of them had attempted to characterize the link from molecular initiating events to a final adverse effect in such a detailed, causality-driven, and formal approach, allowing one to address concerns and limitations of existing hazard assessment from an academic viewpoint. Regulatory testing at present, especially in environmental hazard assessment but to a large extent in the human health sector as well, is mainly based on the analysis of apical endpoints, which rarely provide an understanding of the underlying toxicity mechanism. Lacking mechanistic information, it is difficult to properly judge the hazard of chemicals across different species and exposure situations, derive a sufficient risk assessment, or support a reliable read-across. The assembly and evaluation of AOPs provide a powerful tool to focus research with the purpose of 1) characterizing and assembling the existing knowledge; 2) identifying data gaps (for mechanisms, pathways, biomarkers, assay development, nonanimal models, species susceptibility, etc.); and 3) focusing and directing international research efforts to areas where these may be required. This is supported through data sharing and the recently established AOP-wiki and other knowledge repositories. Furthermore, the AOP concept may foster a stronger link between what is needed from a regulatory perspective, with approaches and technology used in academic research (e.g., toxicogenomics). It will revive earlier concepts such as the use of biomarkers or toxicity pathways by establishing a stronger mechanistic link to adverse effects. Beyond these conceptual advantages, there is already a clear practical usability of the AOP concept in applied academic research. Analysis of apical endpoints such as growth or reproduction requires a large number of test animals and is time-consuming and laborious. Apical endpoint testing may continue to be used for definitive testing of selected compounds, but it is not feasible for testing of a large number of substances. In contrast, molecular initiating events and key events are often detectable using relatively simple in vitro or in chemico approaches or can even be predicted from the chemical structure. If a strong link to the adverse outcome (AO) is established, the toxicity of a chemical might be identified and predicted by measuring and quantifying only the molecular initiating event or key event. This would greatly reduce the number of test animals or even waive (animal) testing and, hence, dramatically reduce the testing load. Furthermore, the AOP concept may be used for comprehensive assessment of mixture effects by relying on AOP networks integrating the information on the connections between multiple key events and AOs 6. This would allow a better characterization of the impact of complex mixtures of contaminants that are typically found in the environment. Finally, in many areas where alternative testing approaches are demanded (e.g., prioritization and high-throughput testing of chemicals or environmental samples), the AOP concept can already be applied without the prior need to conduct validation of the AOP–linked method and comparison with existing approaches. The regulatory community will benefit from academic AOP research because a weight-of-evidence approach based on mechanistic data may facilitate the rapid development and validation of alternative methods. The academic community may be the driver for providing the “weight” to the weight-of-evidence analysis. A further academic perspective for future AOP research, with a strong link to regulatory applications, is to quantitatively describe the causal links of the AOP chain to facilitate the prediction of AOs based on the analysis of molecular initiating events or key events and the extrapolation between models and species. Because the uptake, distribution, biotransformation, and elimination of a chemical determine the bioavailable concentration within an organism and at the target site, toxicokinetic considerations are essential to determine whether a compound at a given exposure concentration will provoke an adverse effect 7, 8. This is even more important when molecular initiating events or key events are measured in in vitro cellular or other screening assays that may exhibit different toxicokinetic properties. For a full quantitative approach, established toxicokinetic and toxicodynamic models could be applied to effectively integrate exposure and toxicokinetic information 9. Such approaches have already been used to explain differences in sensitivity of species 10 and have been suggested in human risk assessment to describe AOPs for skin sensitization 11. At present, however, quantitative approaches have not been practically applied with the AOP concept. The challenge will be to collect, generate, or infer the information necessary for the development and application of quantitative AOPs, such as the concentration and number of targets, affinity to these targets, and concentration time courses of chemicals at the target site. Furthermore, toxicokinetic and exposure data are essential for AOP–informed high-throughput in vitro screening approaches as such information may lead to a prioritization of chemicals that is more relevant to the actual exposure situation for humans and wildlife 8. To demonstrate the strength of the AOP concept for the prediction of adverse effects, it is necessary to identify cases that could be used in proof-of-principle studies. For instance, the AOP for estrogenic receptor binding leading to reproductive impairment 1 and for acetylcholine esterase inhibition leading to acute mortality 12 represent 2 well-described examples that could be used for quantitative and predictive approaches, which are also of high regulatory relevance. Such studies would also show what type of complexity in modeling may finally be required to achieve a certain level of prediction. Stefan Scholz Department of Bioanalytical Ecotoxicology Helmholtz Centre for Environmental Research–UFZ Leipzig, Germany