Quantification of an Adverse Outcome Pathway Network by Bayesian Regression and Bayesian Network Modeling.

S Jannicke Moe,Wayne G Landis,Raoul Wolf,Niina Kotamäki,Li Xie,Knut Erik Tollefsen

doi:10.1002/ieam.4348

Abstract

ABSTRACTThe adverse outcome pathway (AOP) framework has gained international recognition as a systematic approach linking mechanistic processes to toxicity endpoints. Nevertheless, successful implementation into risk assessments is still limited by the lack of quantitative AOP models (qAOPs) and assessment of uncertainties. The few published qAOP models so far are typically based on data‐demanding systems biology models. Here, we propose a less data‐demanding approach for quantification of AOPs and AOP networks, based on regression modeling and Bayesian networks (BNs). We demonstrate this approach with the proposed AOP #245, “Uncoupling of photophosphorylation leading to reduced ATP production associated growth inhibition,” using a small experimental data set from exposure of Lemna minor to the pesticide 3,5‐dichlorophenol. The AOP‐BN reflects the network structure of AOP #245 containing 2 molecular initiating events (MIEs), 3 key events (KEs), and 1 adverse outcome (AO). First, for each dose–response and response–response (KE) relationship, we quantify the causal relationship by Bayesian regression modeling. The regression models correspond to dose–response functions commonly applied in ecotoxicology. Secondly, we apply the fitted regression models with associated uncertainty to simulate 10 000 response values along the predictor gradient. Thirdly, we use the simulated values to parameterize the conditional probability tables of the BN model. The quantified AOP‐BN model can be run in several directions: 1) prognostic inference, run forward from the stressor node to predict the AO level; 2) diagnostic inference, run backward from the AO node; and 3) omnidirectionally, run from the intermediate MIEs and/or KEs. Internal validation shows that the AOP‐BN can obtain a high accuracy rate, when run is from intermediate nodes and when a low resolution is acceptable for the AO. Although the performance of this AOP‐BN is limited by the small data set, our study demonstrates a proof‐of‐concept: the combined use of Bayesian regression modeling and Bayesian network modeling for quantifying AOPs. Integr Environ Assess Manag 2021;17:147–164. © 2020 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC)

Highlights

We will demonstrate how an adverse outcome pathway (AOP)— or even an AOP network—can be quantified on the basis of a small experimental data set, by combining observations with expert knowledge and statistical modeling in a Bayesian framework. This relatively simple approach can serve as a proof‐of‐concept for quantification of AOPs and AOP networks based on limited data sets
AO = adverse outcome; BN = Bayesian network; KE = key event; MIE = molecular initiating event. a The AOP‐BN is instantiated with evidence (Supplemental Data Table S1) for either the stressor node, the MIE nodes, or the KE nodes
Relationship is given by the combination of the up–down arrows in the 2 connected nodes of the conceptual model (Figure 1B): Two nodes with up–down arrows pointing in the same direction represent a positive correlation

Summary

Introduction

An AOP model typically describes the causal linkages from a chemical stressor through 3 types of events: 1) a molecular initiating event (MIE) triggered by the stressor, 2) a series of measurable biological responses termed “key events” (KEs), and 3) 1 or more adverse outcomes (AOs), which are specialized KEs of regulatory significance (Figure 1A). During the last decade there has been a widespread interest and rapid development in the AOP framework by scientists involved in risk assessment both to human health and to the environment (LaLone et al 2017). The AOP Knowledge Base (https://aopkb.oecd.org/) combines all available information on AOP development through 4 different information systems. One of these platforms is the AOPWiki (http://aopwiki.org), which holds descriptions of more than 300 proposed AOPs, with status ranging from “under development” to “adopted by the [Organisation for Economic Co‐operation and Development] OECD.” the AOPWiki is currently dominated by AOPs relevant for human health, the number of AOPs relevant to other animals and plants is increasing, and the number of taxonomic groups is expanding

Methods

Results

Conclusion