Natural language processing in toxicology: Delineating adverse outcome pathways and guiding the application of new approach methodologies

Adverse Outcome Pathways (AOPs) are conceptual frameworks that tie an initial perturbation (molecular initiating event) to a phenotypic toxicological manifestation (adverse outcome), through a series of steps (key events). They provide therefore a standardized way to map and organize toxicological mechanistic information. As such, AOPs inform on key events underlying toxicity, thus supporting the development of New Approach Methodologies (NAMs), which aim to reduce the use of animal testing for toxicology purposes. However, the establishment of a novel AOP relies on the gathering of multiple streams of evidence and information, from available literature to knowledge databases. Often, this information is in the form of free text, also called unstructured text, which is not immediately digestible by a computer. This information is thus both tedious and increasingly time-consuming to process manually with the growing volume of data available. The advancement of machine learning provides alternative solutions to this challenge. To extract and organize information from relevant sources, it seems valuable to employ deep learning Natural Language Processing techniques. We review here some of the recent progress in the NLP field, and show how these techniques have already demonstrated value in the biomedical and toxicology areas. We also propose an approach to efficiently and reliably extract and combine relevant toxicological information from text. This data can be used to map underlying mechanisms that lead to toxicological effects and start building quantitative models, in particular AOPs, ultimately allowing animal-free human-based hazard and risk assessment.

FAIRification of nanosafety data to improve applicability of (Q)SAR approaches: A case study on in vitro Comet assay genotoxicity data

(Quantitative) structure-activity relationship ([Q]SAR) methodologies are widely applied to predict the (eco)toxicological effects of chemicals, and their use is envisaged in different regulatory frameworks for filling data gaps of untested substances. However, their application to the risk assessment of nanomaterials is still limited, also due to the scarcity of large and curated experimental datasets. Despite a great amount of nanosafety data having been produced over the last decade in international collaborative initiatives, their interpretation, integration and reuse has been hampered by several obstacles, such as poorly described (meta)data, non-standard terminology, lack of harmonized reporting formats and criteria. Recently, the FAIR (Findable, Accessible, Interoperable, and Reusable) principles have been established to guide the scientific community in good data management and stewardship. The EU H2020 Gov4Nano project, together with other international projects and initiatives, is addressing the challenge of improving nanosafety data FAIRness, for maximizing their availability, understanding, exchange and ultimately their reuse. These efforts are largely supported by the creation of a common Nanosafety Data Interface, which connects a row of project-specific databases applying the eNanoMapper data model. A wide variety of experimental data relating to characterization and effects of nanomaterials are stored in the database; however, the methods, protocols and parameters driving their generation are not fully mature. This article reports the progress of an ongoing case study in the Gov4nano project on the reuse of in vitro Comet genotoxicity data, focusing on the issues and challenges encountered in their FAIRification through the eNanoMapper data model. The case study is part of an iterative process in which the FAIRification of data supports the understanding of the phenomena underlying their generation and, ultimately, improves their reusability.

Development of a putative adverse outcome pathway network for male rat reproductive tract abnormalities with specific considerations for the androgen sensitive window of development

Structured approaches like the adverse outcome pathway (AOP) framework offer great potential for depicting complex toxicological processes in a manner that can facilitate informed integration of mechanistic information in regulatory decisions. While this concept provides a structure for organizing evidence and facilitates consistency in evidence integration; the process, inputs, and manner in which AOPs and AOP networks are developed is still evolving. Following the OECD guiding principles of AOP development, we propose three AOPs for male reproductive tract abnormalities and derive a putative AOP network. The AOPs were developed using a fundamental understanding of the developmental biology of the organs of interest, paying close attention to the gestational timing of key events (KEs) to very specifically inform the domain of life stage applicability for the key event relationships (KERs). Chemical stressor data primarily from studies on low molecular weight phthalates (LMWPs) served to 'bound' the pathways of focus in this dynamic period of development and were integrated with the developmental biology data through an iterative process to define KEs and conclude on the extent of evidence in support of the KERs. The AOPs developed describe the linkage between 1) a decrease in Insl3 gene expression and cryptorchidism, 2) the sustained expression of Coup-tfII and hypospadias and 3) the sustained expression of Coup-tfII and altered Wolffian duct development/ epididymal agenesis. A putative AOP network linking AOP2 and AOP3 through decreased steroidogenic biosynthetic protein expression and converging of all AOPS at the population level impaired fertility adverse outcome is proposed. The network depiction specifies and displays the KEs aligned with their occurrence in gestational time. The pathways and network described herein are intended to catalyze collaborative initiatives for expansion into a larger network to enable effective data collection and inform alternative approaches for identifying stressors impacting this sensitive period of male reproductive tract development.

COSMOS next generation - A public knowledge base leveraging chemical and biological data to support the regulatory assessment of chemicals

The COSMOS Database (DB) was originally established to provide reliable data for cosmetics-related chemicals within the COSMOS Project funded as part of the SEURAT-1 Research Initiative. The database has subsequently been maintained and developed further into COSMOS Next Generation (NG), a combination of database and in silico tools, essential components of a knowledge base. COSMOS DB provided a cosmetics inventory as well as other regulatory inventories, accompanied by assessment results and in vitro and in vivo toxicity data. In addition to data content curation, much effort was dedicated to data governance - data authorisation, characterisation of quality, documentation of meta information, and control of data use. Through this effort, COSMOS DB was able to merge and fuse data of various types from different sources. Building on the previous effort, the COSMOS Minimum Inclusion (MINIS) criteria for a toxicity database were further expanded to quantify the reliability of studies. COSMOS NG features multiple fingerprints for analysing structure similarity, and new tools to calculate molecular properties and screen chemicals with endpoint-related public profilers, such as DNA and protein binders, liver alerts and genotoxic alerts. The publicly available COSMOS NG enables users to compile information and execute analyses such as category formation and read-across. This paper provides a step-by-step guided workflow for a simple read-across case, starting from a target structure and culminating in an estimation of a NOAEL confidence interval. Given its strong technical foundation, inclusion of quality-reviewed data, and provision of tools designed to facilitate communication between users, COSMOS NG is a first step towards building a toxicological knowledge hub leveraging many public data systems for chemical safety evaluation. We continue to monitor the feedback from the user community at support@mn-am.com.

Multi-endpoint analysis of human 3D airway epithelium following repeated exposure to whole electronic vapor product aerosol or cigarette smoke

Smoking is a cause of serious diseases in smokers including chronic respiratory diseases. This study aimed to evaluate the tobacco harm reduction (THR) potential of an electronic vapor product (EVP, myblu™) compared to a Kentucky Reference Cigarette (3R4F), and assessed endpoints related to chronic respiratory diseases. Endpoints included: cytotoxicity, barrier integrity (TEER), cilia function, immunohistochemistry, and pro-inflammatory markers. In order to more closely represent the user exposure scenario, we have employed the in vitro 3D organotypic model of human airway epithelium (MucilAir™, Epithelix) for respiratory assessment. The model was repeatedly exposed to either whole aerosol of the EVP, or whole 3R4F smoke, at the air liquid interface (ALI), for 4 weeks to either 30, 60 or 90 puffs on 3-exposure-per-week basis. 3R4F smoke generation used the ISO 20778:2018 regime and EVP aerosol used the ISO 20768:2018 vaping regime. Exposure to undiluted whole EVP aerosol did not trigger any significant changes in the level of pro-inflammatory mediators, cilia beating function, barrier integrity and cytotoxicity when compared with air controls. In contrast, exposure to diluted (1:17) whole cigarette smoke caused significant changes to all the endpoints mentioned above. To our knowledge, this is the first study evaluating the effects of repeated whole cigarette smoke and whole EVP aerosol exposure to a 3D lung model at the ALI. Our results add to the growing body of scientific literature supporting the THR potential of EVPs relative to combustible cigarettes and the applicability of the 3D lung models in human-relevant product risk assessments.