Prior knowledge-based approach for associating contaminants with biological effects: A case study in the St. Croix river basin, MN, WI, USA.
Citation:
Schroeder, A., D. Martinović-Weigelt, G. Ankley, K. Lee, N. Garcia-Reyero, E. Perkins, H. Schoenfuss, AND Dan Villeneuve. Prior knowledge-based approach for associating contaminants with biological effects: A case study in the St. Croix river basin, MN, WI, USA. ENVIRONMENTAL POLLUTION. Elsevier Science Ltd, New York, NY, 221:427-436, (2017).
Impact/Purpose:
With the growing access to publically and computationally accessible sources of chemical-biological interaction data in the form of databases of curated scientific literature and/or empirical high throughput screening data there are new opportunities to link chemicals with their known bioactivities, and in turn link bioactivities to hazard through the use of the adverse outcome pathway (AOP) framework. From the standpoint of evaluating risks associated with chemical exposures that are already occurring in the ambient environment these sources create new opportunities for predictive approaches to environmental surveillance, monitoring, and chemical risk assessment. Specifically, application of these sources can be used to both facilitate the inference of potential biological/toxicological hazards associated with chemicals detected in the environment or in organisms and can also be used to help connect observed biological responses with potential causative agents, such as specific chemicals. The present manuscript reports on a case study in which prior knowledge concerning chemical-gene interactions, coded in the Comparative Toxicogenomics Database, were used to develop a “Knowledge Assembly Model” (KAM). The paper demonstrates how a KAM can be used to develop testable hypotheses concerning the potential impacts of chemicals detected in ambient water samples on aquatic biota, in this case fish. Notably, most of chemicals considered are, so called, emerging contaminants for which there are limited apical toxicity data and no established ambient water quality criteria. Further, the case study demonstrates how the relative weight of evidence supporting the idea that any particular chemical detected is contributing to an observed biological response profile, such as alterations in the hepatic transcriptome, can be analyzed statistically to evaluate which contaminants are most likely drivers of biological perturbation. This allows for the integrated use of chemical and biological monitoring data, along with prior knowledge, as a means to prioritize specific chemicals or chemical classes for further monitoring, targeted investigation, and/or source identification and mitigation. With additional development, the use of KAMs, derived from pathway-based data and properly anchored to AOP knowledge, have potential to transform the way that ambient risk assessments are conducted and provide EPAs program offices, Regions, and its partners in state and local government with important new tools for evaluating impacts of emerging contaminants on humans and ecosystems. The present study represents one of the first applications of this type of predictive, pathway-based approach to environmental surveillance and monitoring.
Description:
Evaluating the potential human health and/or ecological risks associated with exposures to complex chemical mixtures in the ambient environment is one of the central challenges of chemical safety assessment and environmental protection. There is a need for approaches that can help to integrate chemical monitoring and bio-effects data to evaluate risks associated with chemicals present in the environment. We used prior knowledge about chemical-gene interactions to develop a knowledge assembly model for detected chemicals at five locations near two wastewater treatment plants. The assembly model was used to generate hypotheses about the biological impacts of the chemicals at each location. The hypotheses were tested using empirical hepatic gene expression data from fathead minnows exposed for 12 d at each location. Empirical gene expression data was also mapped to the assembly models to statistically evaluate the likelihood of a chemical contributing to the observed biological responses. The prior knowledge approach was able reasonably hypothesize the biological impacts at one site but not the other. Chemicals most likely contributing to the observed biological responses were identified at each location. Despite limitations to the approach, knowledge assembly models have strong potential for associating chemical occurrence with potential biological effects and providing a foundation for hypothesis generation to guide research and/or monitoring efforts related to the effects of contaminants in the environment.
