Skip to main content

ADVerse outcome Network for non-mutagenic CarcinogEns

Periodic Reporting for period 1 - ADVaNCE (ADVerse outcome Network for non-mutagenic CarcinogEns)

Reporting period: 2017-10-01 to 2019-09-30

In view of the global legislative changes, which aim to increase the speed of evaluation of chemicals, reinforce the use of alternative methods, and at the same time preserve reliable identification of chemicals of potential risk concern, a better understanding of the mechanisms leading to the development of a specific disease is needed.
The Adverse Outcome Pathway (AOP) is a framework for scientific data organisation, comprised of several building blocks starting with a molecular initiating event (MIE), which describes the initial interaction of a stressor, in most cases a chemical, with a biomolecule and continuing with a dependent series of intermediate key events (KEs) at different levels of biological organisation, while finally ending in an adverse outcome (AO) (Vinken, 2013). The primary objective for the introduction of the AOP framework was knowledge assembly, specifically, making information attained through scientific research by subject-matter experts and distributed in the body of scientific literature accessible to regulators during the decision-making process.

The actual process of AOP development is still mainly done manually, and as such is a slow, time-consuming process with associated problems of sometimes biased and even incorrect events’ linkage.

In this project, we present our advances in the area of computational approaches for AOP hypothesis generation, namely the development of a 2-step automated workflow for AOP hypothesis generation. The created workflow has potential for manifold use and applications in this field, as the incorporation of high-content and high-throughput datasets increasingly becomes a standard part of our toxicological and risk assessment practices.
As a proof-of-principle, we showcase the extraction and identification of KEs at different levels of biological organisation which are linked to non-genotoxic-induced hepatocellular carcinoma (HCC) (a type of liver cancer).

The incorporation of computational toxicology into risk assessment practices is an evolving area. The global goal of this project is to support this expansion by offering a hypothesis generation tool which can be freely exploited by the scientific community and will eventually guide the development of an integrated testing strategy completely based on in silico and in vitro data to be used in regulatory settings. The proposed computationally-predicted AOP has the advantage of being a result of a fast and unbiased process of data gathering, organisation, integration, interlinking and analysis.
The overall objectives are:
-Development of computational approach for pre-processing and linking of information coming from diverse, publicly-available data sources;
-Identification of the best methodologies for creating a collection of all relevant, interlinked events for a particular disease;
-Extraction and identification, grouping and prioritization of (de novo) KEs characteristic for non-mutagenic carcinogen-induced HCC;
-Initiation of a proof-of-principle approach and the in-practice validation of a subset of important KEs of the non-mutagenic AOP network by targeted in vitro experimentation;
-Contribution to the development of an Integrated Testing Strategy (ITS) for identification of non-mutagenic carcinogens.
The main result of the project is an automated workflow for AOP hypothesis generation. In brief, association mining methods were applied to high-throughput screening, gene expression, in vivo and disease data present in ToxCast and the Comparative Toxicogenomics Database. This was supplemented by pathway mapping using Reactome to fill in gaps and identify events occurring at the cellular/tissue levels. Furthermore, in vivo data from TG-Gates was integrated to finally derive a gene, pathway, biochemical, histopathological and disease information network from which specific disease sub-networks can be queried (Figure 1).
To test the workflow, non-genotoxic-induced hepatocellular carcinoma (HCC) was selected as a case study. The implementation resulted in the identification of several non-genotoxic-specific HCC-connected genes belonging to cell proliferation, endoplasmic reticulum stress and early apoptosis. Biochemical findings revealed non-genotoxic-specific alkaline phosphatase increase. The explored non-genotoxic-specific histopathology was associated with early stages of hepatic steatosis, transforming into cirrhosis (Figure 2). As several pathways and intervention points were found as activated in order to contribute to the development of hepatocellular carcinoma, the resulting construct is defined as an AO network (AON) rather than single AOP.

By combining all data coming from diverse data sources to extract biomarkers related to specific KEs and partial AOPs, we demonstrate in practice how to use computational approaches for integration and linking of pre-existing knowledge and how this information combined with expert knowledge can be used in distinguishing between different routes of initiation and manifestation of the same disease. In this way, the work illustrates the utility of computationally predicted constructs in supporting AOP and AON development by using pre-existing knowledge in a fast and unbiased manner.

The results of the project are being described in a manuscript, which is currently under revision (Journal: Regulatory Toxicology and Pharmacology).
Furthermore, during the course of the project the results were presented in a number of conferences and workshops
The main achievement of this project is the creation of a fully automated workflow for data gathering, interlinking and integration. The workflow is building upon data coming from public sources (ToxCast, TG-Gates, Comparative Toxicogenomics Database and Reactome) but has the capacity to integrate also other sources and in the future to become very frequently used by scientists for initial knowledge screening on different diseases or chemicals of interest. The workflow sofar is a series of scripts which are made publicly available for use and adaptation:

The usefulness of the script has been tested for the case of non-genotoxic induced hepatocellular carcinoma and the immediate impact is the development of an AOP network which is showing the major players in the development of the disease as identified by the association mining algorithms and bioinformatics approaches applied in our workflow.

The direct socio-economic impact is increase of the re-use of publicly-available data by application of an unbiased and fast approach. Scientists and risk assessors will spend less time on identification of the mode of action of a new chemical entity or disease of interest which will lead to more focused and unbiased hypotheses to be verified with follow-up testing. In the long run, this has the potential to lead to less use of material and resources and in some cases reduction or replacement of the follow-up animal experimentations, which is often still needed due to uncertainties in the identified mechanisms, their relationships to the adverse effect and quantification of the effect.
Figure 2. The hypothetical non-genotoxic carcinogen AOP construct as identified and gathered as a re
Figure 1. Schematic representation of the two modules of the automated workflow.