Modernization of chemical risk assessment to make use of novel toxicological data

These are challenging and exciting times for environmental science, molecular and computational toxicology. Advances in the biological and molecular sciences have instigated an ongoing paradigm shift in toxicity testing away from traditional, animal-based methods to in vitro and in silico methods to assess the human health risks of environmental agents. High throughput screening efforts accelerate the pace of screening environmental chemicals and continue to reduce animal testing in the US and Europe. Computational efforts to make use of existing data, data standards and data integration issues are currently at the forefront. The application of new approach methodologies is currently in their beginnings.

Importantly, for interpretation of new and existing data streams and the application of these data and methods to human health risk assessment, there is a need for a consistent and structured framework to understand biological process from chemical initiating event to outcome. With the acceptance of the Adverse Outcome Pathway framework ^{1; 2} we are now better positioned to delineate biological mechanism as envisioned by the National Research Council (NRC), while addressing toxicity testing in the 21^st century (TT21C) challenges. The applied usage of the AOP framework for much of computational toxicology focuses on the binning of data types (molecular, cellular, tissue, organ, and population-level) ^{3; 4} and the organization of publicly available data has highlighted data rich areas and data gaps ^{5; 6}. Initial work to organize AOP information ^7–9 underlines the wealth of molecular data, and the use of these types of data to illuminate biological process and mechanism. Elucidation and interpretation of mechanistic effects through case study examples, with a focus on adverse outcome pathway integration, are of specific interest to researchers focused on toxicological effects of environmental chemicals. For molecular, toxicological data, the AOP framework allows for a mapping of data points from molecular initiating events to outcome for putative and computationally derived AOPs ^{10; 11}, with the addition of existing biological pathway and assay information creating a powerful tool for interpretation of biological context and mechanism ^{8; 12}. This has direct implication for the future of integrated testing and risk assessment ^{13; 14}, with specific reference applied to high throughput testing schemes (ToxCast ^{15; 16}; Tox21 ¹⁷).

This special issue on the modernization of chemical risk assessment to make use of molecular toxicological data highlights a temporal slice of the research, mostly government and academic, to understand the applied and integrated sciences of chemical informatics, toxicology, and systems biology, genetics, and computational biology. The focus for this issue is on research that highlights the mechanistic aspects of current toxicological pursuits, in the elucidation of biological mechanism, the use of well-defined mechanism in testing, and the application of mechanistic effects to computational model development and simulation.

In this issue, Corton et al. ¹⁸ describe how we can define biomarkers to identify endocrine disrupting chemicals, and how the same principles can help identify similarly regulated genes with high levels of accuracy for other toxicological outcomes. Kosnik and Reif ¹⁹ address the need to associate chemical risk values to diseases and gene-disease variants in their development of thresholding values to determine human susceptibility. Mezencev and Subramaniam²⁰ provide an excellent review of toxicogenomic applications, in terms of their potential to address the NRC TT21C vision and application to human health risk assessment. Pagé-Larivière et al. ²¹ present a case study example using transcriptomic dose-response analysis in fish to illustrate how estimates of chronic toxicity can be used with high levels of confidence to derive protective points of departure (POD) for endocrine disrupting chemicals (EDCs). Fry et al. ²² review in vitro placental models and associated endpoints that can be used to evaluate chemical-induced toxicity, and in doing so elegantly highlight how these strategies can both screen for hazard and define mechanistic effects of toxicity for developmental adverse outcomes. Abedini et al. ²³ use publicly available data to model molecular processes and illuminate insights into molecular mechanisms for hepatic lipid dysfunction, a precursor for hepatic steatosis. This work is at the foundation of identification of putative key events that are related to lipid homeostasis dysfunction in liver, a possible precursor for the adverse outcome of hepatic steatosis. Watford et al. ²³ focus on the current challenges in data management, interoperability and application of FAIR (findable, accessible, interoperable and reusable) data principles to the integration of NAMs (New Approach Methodologies) to traditional toxicological methods, and present a particularly relevant discussion of how AOP data could be improved through the integration with literature mining tools in order to identify AOPs that are relevant to safety evaluation. And finally, Burnett at al. ²⁴ break ground with their novel population-based cardiotoxicity screening tool.