Integrative chemical-biological read-across approach for chemical hazard classification

A novel multitask learning algorithm for tasks with distinct chemical space: zebrafish toxicity prediction as an example

Data scarcity is one of the most critical issues impeding the development of prediction models for chemical effects. Multitask learning algorithms leveraging knowledge from relevant tasks showed potential for dealing with tasks with limited data. However, current multitask methods mainly focus on learning from datasets whose task labels are available for most of the training samples. Since datasets were generated for different purposes with distinct chemical spaces, the conventional multitask learning methods may not be suitable. This study presents a novel multitask learning method MTForestNet that can deal with data scarcity problems and learn from tasks with distinct chemical space. The MTForestNet consists of nodes of random forest classifiers organized in the form of a progressive network, where each node represents a random forest model learned from a specific task. To demonstrate the effectiveness of the MTForestNet, 48 zebrafish toxicity datasets were collected and utilized as an example. Among them, two tasks are very different from other tasks with only 1.3% common chemicals shared with other tasks. In an independent test, MTForestNet with a high area under the receiver operating characteristic curve (AUC) value of 0.911 provided superior performance over compared single-task and multitask methods. The overall toxicity derived from the developed models of zebrafish toxicity is well correlated with the experimentally determined overall toxicity. In addition, the outputs from the developed models of zebrafish toxicity can be utilized as features to boost the prediction of developmental toxicity. The developed models are effective for predicting zebrafish toxicity and the proposed MTForestNet is expected to be useful for tasks with distinct chemical space that can be applied in other tasks.Scieific contributionA novel multitask learning algorithm MTForestNet was proposed to address the challenges of developing models using datasets with distinct chemical space that is a common issue of cheminformatics tasks. As an example, zebrafish toxicity prediction models were developed using the proposed MTForestNet which provide superior performance over conventional single-task and multitask learning methods. In addition, the developed zebrafish toxicity prediction models can reduce animal testing.

Unlocking the Potential of Clustering and Classification Approaches: Navigating Supervised and Unsupervised Chemical Similarity

BACKGROUND

The field of toxicology has witnessed substantial advancements in recent years, particularly with the adoption of new approach methodologies (NAMs) to understand and predict chemical toxicity. Class-based methods such as clustering and classification are key to NAMs development and application, aiding the understanding of hazard and risk concerns associated with groups of chemicals without additional laboratory work. Advances in computational chemistry, data generation and availability, and machine learning algorithms represent important opportunities for continued improvement of these techniques to optimize their utility for specific regulatory and research purposes. However, due to their intricacy, deep understanding and careful selection are imperative to align the adequate methods with their intended applications.

OBJECTIVES

This commentary aims to deepen the understanding of class-based approaches by elucidating the pivotal role of chemical similarity (structural and biological) in clustering and classification approaches (CCAs). It addresses the dichotomy between general end point-agnostic similarity, often entailing unsupervised analysis, and end point-specific similarity necessitating supervised learning. The goal is to highlight the nuances of these approaches, their applications, and common misuses.

DISCUSSION

Understanding similarity is pivotal in toxicological research involving CCAs. The effectiveness of these approaches depends on the right definition and measure of similarity, which varies based on context and objectives of the study. This choice is influenced by how chemical structures are represented and the respective labels indicating biological activity, if applicable. The distinction between unsupervised clustering and supervised classification methods is vital, requiring the use of end point-agnostic vs. end point-specific similarity definition. Separate use or combination of these methods requires careful consideration to prevent bias and ensure relevance for the goal of the study. Unsupervised methods use end point-agnostic similarity measures to uncover general structural patterns and relationships, aiding hypothesis generation and facilitating exploration of datasets without the need for predefined labels or explicit guidance. Conversely, supervised techniques demand end point-specific similarity to group chemicals into predefined classes or to train classification models, allowing accurate predictions for new chemicals. Misuse can arise when unsupervised methods are applied to end point-specific contexts, like analog selection in read-across, leading to erroneous conclusions. This commentary provides insights into the significance of similarity and its role in supervised classification and unsupervised clustering approaches. https://doi.org/10.1289/EHP14001.

Informing Hazard Identification and Risk Characterization of Environmental Chemicals by Combining Transcriptomic and Functional Data from Human-Induced Pluripotent Stem-Cell-Derived Cardiomyocytes

Environmental chemicals may contribute to the global burden of cardiovascular disease, but experimental data are lacking to determine which substances pose the greatest risk. Human-induced pluripotent stem cell (iPSC)-derived cardiomyocytes are a high-throughput cardiotoxicity model that is widely used to test drugs and chemicals; however, most studies focus on exploring electro-physiological readouts. Gene expression data may provide additional molecular insights to be used for both mechanistic interpretation and dose-response analyses. Therefore, we hypothesized that both transcriptomic and functional data in human iPSC-derived cardiomyocytes may be used as a comprehensive screening tool to identify potential cardiotoxicity hazards and risks of the chemicals. To test this hypothesis, we performed concentration-response analysis of 464 chemicals from 12 classes, including both pharmaceuticals and nonpharmaceutical substances. Functional effects (beat frequency, QT prolongation, and asystole), cytotoxicity, and whole transcriptome response were evaluated. Points of departure were derived from phenotypic and transcriptomic data, and risk characterization was performed. Overall, 244 (53%) substances were active in at least one phenotype; as expected, pharmaceuticals with known cardiac liabilities were the most active. Positive chronotropy was the functional phenotype activated by the largest number of tested chemicals. No chemical class was particularly prone to pose a potential hazard to cardiomyocytes; a varying proportion (10-44%) of substances in each class had effects on cardiomyocytes. Transcriptomic data showed that 69 (15%) substances elicited significant gene expression changes; most perturbed pathways were highly relevant to known key characteristics of human cardiotoxicants. The bioactivity-to-exposure ratios showed that phenotypic- and transcriptomic-based POD led to similar results for risk characterization. Overall, our findings demonstrate how the integrative use of in vitro transcriptomic and phenotypic data from iPSC-derived cardiomyocytes not only offers a complementary approach for hazard and risk prioritization, but also enables mechanistic interpretation of the in vitro test results to increase confidence in decision-making.

Assessing human carcinogenicity risk of agrochemicals without the rodent cancer bioassay

The rodent cancer bioassays are conducted for agrochemical safety assessment yet they often do not inform regulatory decision-making. As part of a collaborative effort, the Rethinking Carcinogenicity Assessment for Agrochemicals Project (ReCAAP) developed a reporting framework to guide a weight of evidence (WOE)-based carcinogenicity assessment that demonstrates how to fulfill the regulatory requirements for chronic risk estimation without the need to conduct lifetime rodent bioassays. The framework is the result of a multi-stakeholder collaboration that worked through an iterative process of writing case studies (in the form of waivers), technical peer reviews of waivers, and an incorporation of key learnings back into the framework to be tested in subsequent case study development. The example waivers used to develop the framework were written retrospectively for registered agrochemical active substances for which the necessary data and information could be obtained through risk assessment documents or data evaluation records from the US EPA. This exercise was critical to the development of a framework, but it lacked authenticity in that the stakeholders reviewing the waiver already knew the outcome of the rodent cancer bioassay(s). Syngenta expanded the evaluation of the ReCAAP reporting framework by writing waivers for three prospective case studies for new active substances where the data packages had not yet been submitted for registration. The prospective waivers followed the established framework considering ADME, potential exposure, subchronic toxicity, genotoxicity, immunosuppression, hormone perturbation, mode of action (MOA), and all relevant information available for read-across using a WOE assessment. The point of departure was estimated from the available data, excluding the cancer bioassay results, with a proposed use for the chronic dietary risk assessment. The read-across assessments compared data from reliable registered chemical analogues to strengthen the prediction of chronic toxicity and/or tumorigenic potential. The prospective case studies represent a range of scenarios, from a new molecule in a well-established chemical class with a known MOA to a molecule with a new pesticidal MOA (pMOA) and limited read-across to related molecules. This effort represents an important step in establishing criteria for a WOE-based carcinogenicity assessment without the rodent cancer bioassay(s) while ensuring a health protective chronic dietary risk assessment.

The pursuit of accurate predictive models of the bioactivity of small molecules

Property prediction is a key interest in chemistry. For several decades there has been a continued and incremental development of mathematical models to predict properties. As more data is generated and accumulated, there seems to be more areas of opportunity to develop models with increased accuracy. The same is true if one considers the large developments in machine and deep learning models. However, along with the same areas of opportunity and development, issues and challenges remain and, with more data, new challenges emerge such as the quality and quantity and reliability of the data, and model reproducibility. Herein, we discuss the status of the accuracy of predictive models and present the authors' perspective of the direction of the field, emphasizing on good practices. We focus on predictive models of bioactive properties of small molecules relevant for drug discovery, agrochemical, food chemistry, natural product research, and related fields.

Advancing Computational Toxicology by Interpretable Machine Learning

Chemical toxicity evaluations for drugs, consumer products, and environmental chemicals have a critical impact on human health. Traditional animal models to evaluate chemical toxicity are expensive, time-consuming, and often fail to detect toxicants in humans. Computational toxicology is a promising alternative approach that utilizes machine learning (ML) and deep learning (DL) techniques to predict the toxicity potentials of chemicals. Although the applications of ML- and DL-based computational models in chemical toxicity predictions are attractive, many toxicity models are "black boxes" in nature and difficult to interpret by toxicologists, which hampers the chemical risk assessments using these models. The recent progress of interpretable ML (IML) in the computer science field meets this urgent need to unveil the underlying toxicity mechanisms and elucidate the domain knowledge of toxicity models. In this review, we focused on the applications of IML in computational toxicology, including toxicity feature data, model interpretation methods, use of knowledge base frameworks in IML development, and recent applications. The challenges and future directions of IML modeling in toxicology are also discussed. We hope this review can encourage efforts in developing interpretable models with new IML algorithms that can assist new chemical assessments by illustrating toxicity mechanisms in humans.

GeoDILI: A Robust and Interpretable Model for Drug-Induced Liver Injury Prediction Using Graph Neural Network-Based Molecular Geometric Representation

Drug-induced liver injury (DILI) is a significant cause of drug failure and withdrawal due to liver damage. Accurate prediction of hepatotoxic compounds is crucial for safe drug development. Several DILI prediction models have been published, but they are built on different data sets, making it difficult to compare model performance. Moreover, most existing models are based on molecular fingerprints or descriptors, neglecting molecular geometric properties and lacking interpretability. To address these limitations, we developed GeoDILI, an interpretable graph neural network that uses a molecular geometric representation. First, we utilized a geometry-based pretrained molecular representation and optimized it on the DILI data set to improve predictive performance. Second, we leveraged gradient information to obtain high-precision atomic-level weights and deduce the dominant substructure. We benchmarked GeoDILI against recently published DILI prediction models, as well as popular GNN models and fingerprint-based machine learning models using the same data set, showing superior predictive performance of our proposed model. We applied the interpretable method in the DILI data set and derived seven precise and mechanistically elucidated structural alerts. Overall, GeoDILI provides a promising approach for accurate and interpretable DILI prediction with potential applications in drug discovery and safety assessment. The data and source code are available at GitHub repository (https://github.com/CSU-QJY/GeoDILI).

Investigation of chemical structure recognition by encoder-decoder models in learning progress

Descriptor generation methods using latent representations of encoder-decoder (ED) models with SMILES as input are useful because of the continuity of descriptor and restorability to the structure. However, it is not clear how the structure is recognized in the learning progress of ED models. In this work, we created ED models of various learning progress and investigated the relationship between structural information and learning progress. We showed that compound substructures were learned early in ED models by monitoring the accuracy of downstream tasks and input-output substructure similarity using substructure-based descriptors, which suggests that existing evaluation methods based on the accuracy of downstream tasks may not be sensitive enough to evaluate the performance of ED models with SMILES as descriptor generation methods. On the other hand, we showed that structure restoration was time-consuming, and in particular, insufficient learning led to the estimation of a larger structure than the actual one. It can be inferred that determining the endpoint of the structure is a difficult task for the model. To our knowledge, this is the first study to link the learning progress of SMILES by ED model to chemical structures for a wide range of chemicals.

Computer-Aided drug design of new 2-amino-thiophene derivatives as anti-leishmanial agents

The leishmaniasis is a neglected disease caused by a group of protozoan parasites from the genus Leishmania whose treatment is limited, obsolete, toxic, and ineffective in certain cases. These characteristics motivate researchers worldwide to plan new therapeutic alternatives for the treatment of leishmaniasis, where the use of cheminformatics tools applied to computer-assisted drug design has allowed research to make great advances in the search for new drugs candidates. In this study, a series of 2-amino-thiophene (2-AT) derivatives was screened virtually using QSAR tools, ADMET filters and prediction models, allowing direct the synthesis of compounds, which were evaluated in vitro against promastigotes and axenic amastigotes of Leishmania amazonensis. The combination of different descriptors and machine learning methods led to obtaining robust and predictive QSAR models, which was obtained from a dataset composed of 1862 compounds extracted from the ChEMBL database, with correct classification rates ranging from 0.53 (for amastigotes) to 0.91 (for promastigotes), allowing to select eleven 2-AT derivatives, which do not violate Lipinski's rules, exhibit good druglikeness, and with probability ≤70% of potential activity against the two evolutionary forms of the parasite. All compounds were properly synthesized and 8 of them were shown to be active at least against one of the evolutionary forms of the parasite with IC₅₀ values lower than 10 μM, being more active than the reference drug meglumine antimoniate, and showing low or no citotoxicity against macrophage J774.A1 for the most part. Compounds 8CN and DCN-83, respectively, are the most active against promastigote and amastigote forms, with IC₅₀ values of 1.20 and 0.71 μM, and selectivity indexes (SI) of 36.58 and 119.33. Structure Activity Relationship (SAR) study was carried out and allowed to identify some favorable and/or essential substitution patterns for the leishmanial activity of 2-AT derivatives. Taken together, these findings demonstrate that the use of ligand-based virtual screening proved to be quite effective and saved time, effort, and money in the selection of potential anti-leishmanial agents, and confirm, once again that 2-AT derivatives are promising hit compounds for the development of new anti-leishmanial agents.

Significant shifts in preclinical and clinical neurotoxicology: a review and commentary

The ever-expanding prevalence of adverse neurotoxic reactions of the brain in response to therapeutic and recreational drugs, dietary supplements, environmental hazards, cosmetic ingredients, a spectrum of herbals, health status, and environmental stressors continues to prompt the development of novel cell-based assays to better determine neurotoxic hazard. Neurotoxicants may cause direct and epigenetic damage to the nervous tissue and alter the chemistry, structure, or normal activity of the nervous system. In severe neurotoxicity due to exposure to physical or psychosocial toxicants, neurons are disrupted or killed, and a consistent pattern of clinical neural dysfunction appears. In utero exposure to neurotoxicants can lead to altered development of the nervous system [developmental neurotoxicity (DNT)]. Patients with certain disorders and certain genomic makeup may be particularly susceptible to neurotoxicants. Traditional cytotoxicity measurements, like cell death, are easy to measure, but insufficient at identifying current routine biomarkers of toxicity including functional impairment in cell communication, which often occurs before or even in the absence of cell death. The present paper examines some of the limitations of existing neurotoxicology in light of the increasing need to develop tools to meet the challenges of achieving greater sensitivity in detection and developing and standardizing methods for exploring the toxicologic risk of such neurotoxic entities as engineered nanomaterials and even variables associated with poverty.

Quantifying Analogue Suitability for SAR-Based Read-Across Toxicological Assessment

Structure activity relationship (SAR)-based read-across often is an integral part of toxicological safety assessment, and justification of the prediction presents the most challenging aspect of the approach. It has been established that structural consideration alone is inadequate for selecting analogues and justifying their use, and biological relevance must be incorporated. Here we introduce an approach for considering biological and toxicological related features quantitatively to compute a similarity score that is concordant with suitability for a read-across prediction for systemic toxicity. Fingerprint keys for comparing metabolism, reactivity, and physical chemical properties are presented and used to compare these attributes for 14 case study chemicals each with a list of potential analogues. Within each case study, the sum of these nonstructural similarity scores is consistent with suitability for read-across established using an approach based on expert judgment. Machine learning is applied to determine the contributions from each of the similarity attributes revealing their importance for each structure class. This approach is used to quantify and communicate the differences between a target and a potential analogue as well as rank analogue quality when more than one is relevant. A numerical score with easily interpreted fingerprints increases transparency and consistency among experts, facilitates implementation by others, and ultimately increases chances for regulatory acceptance.

Navigating Transcriptomic Connectivity Mapping Workflows to Link Chemicals with Bioactivities

Screening new compounds for potential bioactivities against cellular targets is vital for drug discovery and chemical safety. Transcriptomics offers an efficient approach for assessing global gene expression changes, but interpreting chemical mechanisms from these data is often challenging. Connectivity mapping is a potential data-driven avenue for linking chemicals to mechanisms based on the observation that many biological processes are associated with unique gene expression signatures (gene signatures). However, mining the effects of a chemical on gene signatures for biological mechanisms is challenging because transcriptomic data contain thousands of noisy genes. New connectivity mapping approaches seeking to distinguish signal from noise continue to be developed, spurred by the promise of discovering chemical mechanisms, new drugs, and disease targets from burgeoning transcriptomic data. Here, we analyze these approaches in terms of diverse transcriptomic technologies, public databases, gene signatures, pattern-matching algorithms, and statistical evaluation criteria. To navigate the complexity of connectivity mapping, we propose a harmonized scheme to coherently organize and compare published workflows. We first standardize concepts underlying transcriptomic profiles and gene signatures based on various transcriptomic technologies such as microarrays, RNA-Seq, and L1000 and discuss the widely used data sources such as Gene Expression Omnibus, ArrayExpress, and MSigDB. Next, we generalize connectivity mapping as a pattern-matching task for finding similarity between a query (e.g., transcriptomic profile for new chemical) and a reference (e.g., gene signature of known target). Published pattern-matching approaches fall into two main categories: vector-based use metrics like correlation, Jaccard index, etc., and aggregation-based use parametric and nonparametric statistics (e.g., gene set enrichment analysis). The statistical methods for evaluating the performance of different approaches are described, along with comparisons reported in the literature on benchmark transcriptomic data sets. Lastly, we review connectivity mapping applications in toxicology and offer guidance on evaluating chemical-induced toxicity with concentration-response transcriptomic data. In addition to serving as a high-level guide and tutorial for understanding and implementing connectivity mapping workflows, we hope this review will stimulate new algorithms for evaluating chemical safety and drug discovery using transcriptomic data.

Advancing the science of a read-across framework for evaluation of data-poor chemicals incorporating systematic and new approach methods

The assessment of human health hazards posed by chemicals traditionally relies on toxicity studies in experimental animals. However, most chemicals currently in commerce do not meet the minimum data requirements for hazard identification and dose-response analysis in human health risk assessment. Previously, we introduced a read-across framework designed to address data gaps for screening-level assessment of chemicals with insufficient in vivo toxicity information (Wang et al., 2012). It relies on inference by analogy from suitably tested source analogues to a target chemical, based on structural, toxicokinetic, and toxicodynamic similarity. This approach has been used for dose-response assessment of data-poor chemicals relevant to the U.S. EPA's Superfund program. We present herein, case studies of the application of this framework, highlighting specific examples of the use of biological similarity for chemical grouping and quantitative read-across. Based on practical knowledge and technological advances in the fields of read-across and predictive toxicology, we propose a revised framework. It includes important considerations for problem formulation, systematic review, target chemical analysis, analogue identification, analogue evaluation, and incorporation of new approach methods. This work emphasizes the integration of systematic methods and alternative toxicity testing data and tools in chemical risk assessment to inform regulatory decision-making.

Evaluating structure-based activity in a high-throughput assay for steroid biosynthesis

Data from a high-throughput human adrenocortical carcinoma assay (HT-H295R) for steroid hormone biosynthesis are available for >2000 chemicals in single concentration and 654 chemicals in multi-concentration (mc). Previously, a metric describing the effect size of a chemical on the biosynthesis of 11 hormones was derived using mc data referred to as the maximum mean Mahalanobis distance (maxmMd). However, mc HT-H295R assay data remain unavailable for many chemicals. This work leverages existing HT-H295R assay data by constructing structure-activity relationships to make predictions for data-poor chemicals, including: (1) identification of individual structural descriptors, known as ToxPrint chemotypes, associated with increased odds of affecting estrogen or androgen synthesis; (2) a random forest (RF) classifier using physicochemical property descriptors to predict HT-H295R maxmMd binary (positive or negative) outcomes; and, (3) a local approach to predict maxmMd binary outcomes using nearest neighbors (NNs) based on two types of chemical fingerprints (chemotype or Morgan). Individual chemotypes demonstrated high specificity (85-98%) for modulators of estrogen and androgen synthesis but with low sensitivity. The best RF model for maxmMd classification included 13 predicted physicochemical descriptors, yielding a balanced accuracy (BA) of 71% with only modest improvement when hundreds of structural features were added. The best two NN models for binary maxmMd prediction demonstrated BAs of 85 and 81% using chemotype and Morgan fingerprints, respectively. Using an external test set of 6302 chemicals (lacking HT-H295R data), 1241 were identified as putative estrogen and androgen modulators. Combined results across the three classification models (global RF model and two local NN models) predict that 1033 of the 6302 chemicals would be more likely to affect HT-H295R bioactivity. Together, these in silico approaches can efficiently prioritize thousands of untested chemicals for screening to further evaluate their effects on steroid biosynthesis.

Natural Products from Annonaceae as Potential Antichagasic Agents

Chagas disease, a neglected tropical disease, is endemic in 21 Latin American countries and particularly prevalent in Brazil. Chagas disease has drawn more attention in recent years due to its expansion into non-endemic areas. The aim of this work was to computationally identify and experimentally validate the natural products from an Annonaceae family as antichagasic agents. Through the ligand-based virtual screening, we identified 57 molecules with potential activity against the epimastigote form of T. cruzi. Then, 16 molecules were analyzed in the in vitro study, of which, six molecules displayed previously unknown antiepimastigote activity. We also evaluated these six molecules for trypanocidal activity. We observed that all six molecules have potential activity against the amastigote form, but no molecules were active against the trypomastigote form. 13-Epicupressic acid seems to be the most promising, as it was predicted as an active compound in the in silico study against the amastigote form of T. cruzi, in addition to having in vitro activity against the epimastigote form.

Computer-Assisted Discovery of Alkaloids with Schistosomicidal Activity

Schistosomiasis is a chronic parasitic disease caused by trematodes of the genus Schistosoma; it is commonly caused by Schistosoma mansoni, which is transmitted by Bioamphalaria snails. Studies show that more than 200 million people are infected and that more than 90% of them live in Africa. Treatment with praziquantel has the best cost–benefit result on the market. However, hypersensitivity, allergy, and drug resistance are frequently presented after administration. From this perspective, ligand-based and structure-based virtual screening (VS) techniques were combined to select potentially active alkaloids against S. mansoni from an internal dataset (SistematX). A set of molecules with known activity against S. mansoni was selected from the ChEMBL database to create two different models with accuracy greater than 84%, enabling ligand-based VS of the alkaloid bank. Subsequently, structure-based VS was performed through molecular docking using four targets of the parasite. Finally, five consensus hits (i.e., five alkaloids with schistosomicidal potential), were selected. In addition, in silico evaluations of the metabolism, toxicity, and drug-like profile of these five selected alkaloids were carried out. Two of them, namely, 11,12-methylethylenedioxypropoxy and methyl-3-oxo-12-methoxy-n(1)-decarbomethoxy-14,15-didehydrochanofruticosinate, had plausible toxicity, metabolomics, and toxicity profiles. These two alkaloids could serve as starting points for the development of new schistosomicidal compounds based on natural products.

A 10-step framework for use of read-across (RAX) in next generation risk assessment (NGRA) for cosmetics safety assessment

This paper presents a 10-step read-across (RAX) framework for use in cases where a threshold of toxicological concern (TTC) approach to cosmetics safety assessment is not possible. RAX builds on established approaches that have existed for more than two decades using chemical properties and in silico toxicology predictions, by further substantiating hypotheses on toxicological similarity of substances, and integrating new approach methodologies (NAM) in the biological and kinetic domains. NAM include new types of data on biological observations from, for example, in vitro assays, toxicogenomics, metabolomics, receptor binding screens and uses physiologically-based kinetic (PBK) modelling to inform about systemic exposure. NAM data can help to substantiate a mode/mechanism of action (MoA), and if similar chemicals can be shown to work by a similar MoA, a next generation risk assessment (NGRA) may be performed with acceptable confidence for a data-poor target substance with no or inadequate safety data, based on RAX approaches using data-rich analogue(s), and taking account of potency or kinetic/dynamic differences.

In Silico Models for Predicting Acute Systemic Toxicity

In this chapter, we give a brief overview of the regulatory requirements for acute systemic toxicity information in the European Union, and we review structure-based computational models that are available and potentially useful in the assessment of acute systemic toxicity. Emphasis is placed on quantitative structure-activity relationship (QSAR) models implemented by means of a range of software tools. The most recently published literature models for acute systemic toxicity are also discussed, and perspectives for future developments in this field are offered.

OUP accepted manuscript

BlueScreen HC is a mammalian cell-based assay for measuring the genotoxicity and cytotoxicity of chemical compounds and mixtures. The BlueScreen HC assay has been utilized at the Research Institute for Fragrance Materials in a safety assessment program as a screening tool to prioritize fragrance materials for higher-tier testing, as supporting evidence when using a read-across approach, and as evidence to adjust the threshold of toxicological concern. Predictive values for the BlueScreen HC assay were evaluated based on the ability of the assay to predict the outcome of in vitro and in vivo mutagenicity and chromosomal damage genotoxicity assays. A set of 371 fragrance materials was assessed in the BlueScreen HC assay along with existing or newly generated in vitro and in vivo genotoxicity data. Based on a weight-of-evidence approach, the majority of materials in the data set were deemed negative and concluded not to have the potential to be genotoxic, while only a small proportion of materials were determined to show genotoxic effects in these assays. Analysis of the data set showed a combination of high positive agreement but low negative agreement between BlueScreen HC results, in vitro regulatory genotoxicity assays, and higher-tier test results. The BlueScreen HC assay did not generate any false negatives, thereby providing robustness when utilizing it as a high-throughput screening tool to evaluate the large inventory of fragrance materials. From the perspective of protecting public health, it is desirable to have no or minimal false negatives, as a false-negative result may incorrectly indicate the lack of a genotoxicity hazard. However, the assay did have a high percentage of false-positive results, resulting in poor positive predictivity of the in vitro genotoxicity test battery outcome. Overall, the assay generated 100% negative predictivity and 3.9% positive predictivity. In addition to the data set of 371 fragrance materials, 30 natural complex substances were evaluated for BlueScreen HC, Ames, and in vitro micronucleus assay, and a good correlation in all three assays was observed. Overall, while a positive result may have to be further investigated, these findings suggest that the BlueScreen HC assay can be a valuable screening tool to detect the genotoxic potential of fragrance materials and mixtures.

CurveP Method for Rendering High-Throughput Screening Dose-Response Data into Digital Fingerprints

The nature of high-throughput screening (HTS) puts certain limits on optimal test conditions for each particular sample; therefore, on top of usual data normalization, additional parsing is often needed to account for incomplete read outs or various artifacts that arise from signal interferences.CurveP is a heuristic, user-tunable curve-cleaning algorithm that attempts to find a minimum set of corrections, which would give a monotonic dose-response curve. After applying the corrections, the algorithm proceeds to calculate a set of numeric features, which can be used as a fingerprint characterizing the sample, or as a vector of independent variables (e.g., molecular descriptors in case of chemical substances testing). The resulting output can be a part of HTS data analysis or can be used as input for a broad spectrum of computational applications, such as quantitative structure-activity relationship (QSAR ) modeling, computational toxicology, bioinformatics, and cheminformatics.

A read-across case study on chronic toxicity of branched carboxylic acids (1): Integration of mechanistic evidence from new approach methodologies (NAMs) to explore a common mode of action

This read-across case study characterises thirteen, structurally similar carboxylic acids demonstrating the application of in vitro and in silico human-based new approach methods, to determine biological similarity. Based on data from in vivo animal studies, the read-across hypothesis is that all analogues are steatotic and so should be considered hazardous. Transcriptomic analysis to determine differentially expressed genes (DEGs) in hepatocytes served as first tier testing to confirm a common mode-of-action and identify differences in the potency of the analogues. An adverse outcome pathway (AOP) network for hepatic steatosis, informed the design of an in vitro testing battery, targeting AOP relevant MIEs and KEs, and Dempster-Shafer decision theory was used to systematically quantify uncertainty and to define the minimal testing scope. The case study shows that the read-across hypothesis is the critical core to designing a robust, NAM-based testing strategy. By summarising the current mechanistic understanding, an AOP enables the selection of NAMs covering MIEs, early KEs, and late KEs. Experimental coverage of the AOP in this way is vital since MIEs and early KEs alone are not confirmatory of progression to the AO. This strategy exemplifies the workflow previously published by the EUTOXRISK project driving a paradigm shift towards NAM-based NGRA.

Repeat-dose toxicity prediction with Generalized Read-Across (GenRA) using targeted transcriptomic data: A proof-of-concept case study

Read-across is a data gap filling technique utilized to predict the toxicity of a target chemical using data from similar analogues. Recent efforts such as Generalized Read-Across (GenRA) facilitate automated read-across predictions for untested chemicals. GenRA makes predictions of toxicity outcomes based on "neighboring" chemicals characterized by chemical and bioactivity fingerprints. Here we investigated the impact of biological similarities on neighborhood formation and read-across performance in predicting hazard (based on repeat-dose testing outcomes from US EPA ToxRefDB v2.0). We used targeted transcriptomic data on 93 genes for 1060 chemicals in HepaRG^™ cells that measure nuclear receptor activation, xenobiotic metabolism, cellular stress, cell cycle progression, and apoptosis. Transcriptomic similarity between chemicals was calculated using binary hit-calls from concentration-response data for each gene. We evaluated GenRA performance in predicting ToxRefDB v2.0 hazard outcomes using the area under the Receiver Operating Characteristic (ROC) curve (AUC) for the baseline approach (chemical fingerprints) versus transcriptomic fingerprints and a combination of both (hybrid). For all endpoints, there were significant but only modest improvements in ROC AUC scores of 0.01 (2.1%) and 0.04 (7.3%) with transcriptomic and hybrid descriptors, respectively. However, for liver-specific toxicity endpoints, ROC AUC scores improved by 10% and 17% for transcriptomic and hybrid descriptors, respectively. Our findings suggest that using hybrid descriptors formed by combining chemical and targeted transcriptomic information can improve in vivo toxicity predictions in the right context.

Progress towards an OECD reporting framework for transcriptomics and metabolomics in regulatory toxicology

Omics methodologies are widely used in toxicological research to understand modes and mechanisms of toxicity. Increasingly, these methodologies are being applied to questions of regulatory interest such as molecular point-of-departure derivation and chemical grouping/read-across. Despite its value, widespread regulatory acceptance of omics data has not yet occurred. Barriers to the routine application of omics data in regulatory decision making have been: 1) lack of transparency for data processing methods used to convert raw data into an interpretable list of observations; and 2) lack of standardization in reporting to ensure that omics data, associated metadata and the methodologies used to generate results are available for review by stakeholders, including regulators. Thus, in 2017, the Organisation for Economic Co-operation and Development (OECD) Extended Advisory Group on Molecular Screening and Toxicogenomics (EAGMST) launched a project to develop guidance for the reporting of omics data aimed at fostering further regulatory use. Here, we report on the ongoing development of the first formal reporting framework describing the processing and analysis of both transcriptomic and metabolomic data for regulatory toxicology. We introduce the modular structure, content, harmonization and strategy for trialling this reporting framework prior to its publication by the OECD.

Application of the 3Rs principles in the development of pharmaceutical generics

Although the 3Rs are broadly applied in nonclinical testing, a better appreciation of the 3Rs is needed in the field of differentiated or value-added pharmaceutical generics because the minor changes in formulation, dosage form, indication, and application route often do not require additional safety testing. The US FDA and the EU EMA have comprehensive regulations for such drugs based on quality, therapeutic equivalence, and safety guidelines. However, no scientific publications on how the concept of replacement and reduction from 3Rs principles can be applied in the safety assessment of differentiated generics were found in the public domain. In this review, we discuss the application of 3Rs in nonclinical testing requirements for differentiated generics. Practical examples are provided in the form of case studies from regulated markets. We highlight the need for utilization of existing data to establish equivalence (differentiated generic vs innovator) in efficacy and safety. The case studies indicate that data requirements from animal experiments have been reduced to a large extent in some major markets without compromising quality and safety. In this context, we also highlight the problem that on a global scale, a true reduction of animal experiments will only be achieved when all countries adopt similar practices.

Mechanism-Driven Read-Across of Chemical Hepatotoxicants Based on Chemical Structures and Biological Data

Hepatotoxicity is a leading cause of attrition in the drug development process. Traditional preclinical and clinical studies to evaluate hepatotoxicity liabilities are expensive and time consuming. With the advent of critical advancements in high-throughput screening, there has been a rapid accumulation of in vitro toxicity data available to inform the risk assessment of new pharmaceuticals and chemicals. To this end, we curated and merged all available in vivo hepatotoxicity data obtained from the literature and public resources, which yielded a comprehensive database of 4089 compounds that includes hepatotoxicity classifications. After dividing the original database of chemicals into modeling and test sets, PubChem assay data were automatically extracted using an in-house data mining tool and clustered based on relationships between structural fragments and cellular responses in in vitro assays. The resultant PubChem assay clusters were further investigated. During the cross-validation procedure, the biological data obtained from several assay clusters exhibited high predictivity of hepatotoxicity and these assays were selected to evaluate the test set compounds. The read-across results indicated that if a new compound contained specific identified chemical fragments (ie, Molecular Initiating Event) and showed active responses in the relevant selected PubChem assays, there was potential for the chemical to be hepatotoxic in vivo. Furthermore, several mechanisms that might contribute to toxicity were derived from the modeling results including alterations in nuclear receptor signaling and inhibition of DNA repair. This modeling strategy can be further applied to the investigation of other complex chemical toxicity phenomena (eg, developmental and reproductive toxicities) as well as drug efficacy.

Generalised Read-Across prediction using genra-py

MOTIVATION

Generalised Read-Across (GenRA) is a data-driven approach to estimate physico-chemical, biological, or eco-toxicological properties of chemicals by inference from analogues. GenRA attempts to mimic a human expert's manual read-across reasoning for filling data gaps about new chemicals from known chemicals with an interpretable and automated approach based on nearest-neighbours. A key objective of GenRA is to systematically explore different choices of input data selection and neighbourhood definition to objectively evaluate predictive performance of automated read-across estimates of chemical properties.

RESULTS

We have implemented genra-py as a python package that can be freely used for chemical safety analysis and risk assessment applications. Automated read-across prediction in genra-py conforms to the scikit-learn machine learning library's estimator design pattern, making it easy to use and integrate in computational pipelines. We demonstrate the data-driven application of genra-py to address two key human health risk assessment problems namely: hazard identification and point of departure estimation.

AVAILABILITY

The package is available from github.com/i-shah/genra-py.

Reliable Surface Analysis Data of Nanomaterials in Support of Risk Assessment Based on Minimum Information Requirements

The minimum information requirements needed to guarantee high-quality surface analysis data of nanomaterials are described with the aim to provide reliable and traceable information about size, shape, elemental composition and surface chemistry for risk assessment approaches. The widespread surface analysis methods electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS) and secondary ion mass spectrometry (SIMS) were considered. The complete analysis sequence from sample preparation, over measurements, to data analysis and data format for reporting and archiving is outlined. All selected methods are used in surface analysis since many years so that many aspects of the analysis (including (meta)data formats) are already standardized. As a practical analysis use case, two coated TiO2 reference nanoparticulate samples, which are available on the Joint Research Centre (JRC) repository, were selected. The added value of the complementary analysis is highlighted based on the minimum information requirements, which are well-defined for the analysis methods selected. The present paper is supposed to serve primarily as a source of understanding of the high standardization level already available for the high-quality data in surface analysis of nanomaterials as reliable input for the nanosafety community.

Application of grouping and read-across for the evaluation of parabens of different chain lengths with a particular focus on endocrine properties

This article presents the outcomes of higher-tier repeated-dose toxicity studies and developmental and reproductive toxicity (DART) studies using Wistar rats requested for methyl paraben and propyl paraben under the European Union chemicals legislation. All studies revealed no-observed adverse effects (NOAELs) at 1000 mg/kg body weight/day. These findings (absence of effects) were then used to interpolate the hazard profile for ethyl paraben, further considering available data for butyl paraben. The underlying read-across hypothesis (all shorter-chained linear n-alkyl parabens are a ‘category’ based on very high structural similarity and are transformed to a common compound) was confirmed by similarity calculations and comparative in vivo toxicokinetics screening studies for methyl paraben, ethyl paraben, propyl paraben and butyl paraben. All four parabens were rapidly taken up systemically following oral gavage administration to rats, metabolised to p-hydroxybenzoic acid, and rapidly eliminated (parabens within one hour; p-hydroxybenzoic acid within 4–8 h). Accordingly, for ethyl paraben, the NOAELs for repeated-dose toxicity and DART were interpolated to be 1000 mg/kg body weight/day. Finally, all evidence was evaluated to address concerns expressed in the literature that parabens might be endocrine disruptors. This evaluation showed that the higher-tier studies do not provide any indication for any endocrine disrupting property. This is the first time that a comprehensive dataset from higher-tier in vivo studies following internationally agreed test protocols has become available for shorter-chained linear n-alkyl parabens. Consistently, the dataset shows that these parabens are devoid of repeated-dose toxicity and do not possess any DART or endocrine disrupting properties.

Ligand and Structure-based Virtual Screening of Lamiaceae Diterpenes with Potential Activity against a Novel Coronavirus (2019-nCoV)

BACKGROUND

The emergence of a new coronavirus (CoV), named 2019-nCoV, as an outbreak originated in the city of Wuhan, China, has resulted in the death of more than 3,400 people this year alone and has caused worldwide an alarming situation, particularly following previous CoV epidemics, including the Severe Acute Respiratory Syndrome (SARS) in 2003 and the Middle East Respiratory Syndrome (MERS) in 2012. Currently, no exists for infections caused by CoVs; however, some natural products may represent potential treatment resources, such as those that contain diterpenes.

OBJECTIVE

This study aimed to use computational methods to perform a virtual screening (VS) of candidate diterpenes with the potential to act as CoV inhibitors.

METHODS

1,955 diterpenes, derived from the Nepetoideae subfamily (Lamiaceae), were selected using the SistematX tool (https://sistematx.ufpb.br), which were used to make predictions. From the ChEMBL database, 3 sets of chemical structures were selected for the construction of predictive models.

RESULTS

The chemical structures of molecules with known activity against SARS CoV, two of which were tested for activity against specific viral proteins and one of which was tested for activity against the virus itself, were classified according to their pIC50 values [-log IC50 (mol/l)].

CONCLUSION

In the consensus analysis approach, combining both ligand- and structure-based VSs, 19 compounds were selected as potential CoV inhibitors, including isotanshinone IIA (01), tanshinlactone (02), isocryptotanshinone (03), and tanshinketolactone (04), which did not present toxicity within the evaluated parameters.

Grouping of UVCB substances with new approach methodologies (NAMs) data

One of the most challenging areas in regulatory science is assessment of the substances known as UVCB (unknown or variable composition, complex reaction products and biological materials). Because the inherent complexity and variability of UVCBs present considerable challenges for establishing sufficient substance similarity based on chemical characteristics or other data, we hypothesized that new approach methodologies (NAMs), including in vitro test-derived biological activity signatures to characterize substance similarity, could be used to support grouping of UVCBs. We tested 141 petroleum substances as representative UVCBs in a compendium of 15 human cell types representing a variety of tissues. Petroleum substances were assayed in dilution series to derive point of departure estimates for each cell type and phenotype. Extensive quality control measures were taken to ensure that only high-confidence in vitro data were used to determine whether current groupings of these petroleum substances, based largely on the manufacturing process and physico-chemical properties, are justifiable. We found that bioactivity data-based groupings of petroleum substances were generally consistent with the manufacturing class-based categories. We also showed that these data, especially bioactivity from human induced pluripotent stem cell (iPSC)-derived and primary cells, can be used to rank substances in a manner highly concordant with their expected in vivo hazard potential based on their chemical compositional profile. Overall, this study demonstrates that NAMs can be used to inform groupings of UVCBs, to assist in identification of representative substances in each group for testing when needed, and to fill data gaps by read-across.

Variability in in vivo studies: Defining the upper limit of performance for predictions of systemic effect levels

New approach methodologies (NAMs) for chemical hazard assessment are often evaluated via comparison to animal studies; however, variability in animal study data limits NAM accuracy. The US EPA Toxicity Reference Database (ToxRefDB) enables consideration of variability in effect levels, including the lowest effect level (LEL) for a treatment-related effect and the lowest observable adverse effect level (LOAEL) defined by expert review, from subacute, subchronic, chronic, multi-generation reproductive, and developmental toxicity studies. The objectives of this work were to quantify the variance within systemic LEL and LOAEL values, defined as potency values for effects in adult or parental animals only, and to estimate the upper limit of NAM prediction accuracy. Multiple linear regression (MLR) and augmented cell means (ACM) models were used to quantify the total variance, and the fraction of variance in systemic LEL and LOAEL values explained by available study descriptors (e.g., administration route, study type). The MLR approach considered each study descriptor as an independent contributor to variance, whereas the ACM approach combined categorical descriptors into cells to define replicates. Using these approaches, total variance in systemic LEL and LOAEL values (in log10-mg/kg/day units) ranged from 0.74 to 0.92. Unexplained variance in LEL and LOAEL values, approximated by the residual mean square error (MSE), ranged from 0.20-0.39. Considering subchronic, chronic, or developmental study designs separately resulted in similar values. Based on the relationship between MSE and R-squared for goodness-of-fit, the maximal R-squared may approach 55 to 73% for a NAM-based predictive model of systemic toxicity using these data as reference. The root mean square error (RMSE) ranged from 0.47 to 0.63 log10-mg/kg/day, depending on dataset and regression approach, suggesting that a two-sided minimum prediction interval for systemic effect levels may have a width of 58 to 284-fold. These findings suggest quantitative considerations for building scientific confidence in NAM-based systemic toxicity predictions.

Cardiovascular Effects of Polychlorinated Biphenyls and Their Major Metabolites

BACKGROUND

Xenobiotic metabolism is complex, and accounting for bioactivation and detoxification processes of chemicals remains among the most challenging aspects for decision making with in vitro new approach methods data.

OBJECTIVES

Considering the physiological relevance of human organotypic culture models and their utility for high-throughput screening, we hypothesized that multidimensional chemical-biological profiling of chemicals and their major metabolites is a sensible alternative for the toxicological characterization of parent molecules vs. metabolites in vitro.

METHODS

In this study, we tested 25 polychlorinated biphenyls (PCBs) [PCB 3, 11, 52, 126, 136, and 153 and their relevant metabolites (hydroxylated, methoxylated, sulfated, and quinone)] in concentration-response (10 nM-100μM) for effects in human induced pluripotent stem cell (iPSC)-derived cardiomyocytes (CMs) and endothelial cells (ECs) (iPSC-derived and HUVECs). Functional phenotypic end points included effects on beating parameters and intracellular Ca2+ flux in CMs and inhibition of tubulogenesis in ECs. High-content imaging was used to evaluate cytotoxicity, mitochondrial integrity, and oxidative stress.

RESULTS

Data integration of a total of 19 physicochemical descriptors and 36 in vitro phenotypes revealed that chlorination status and metabolite class are strong predictors of the in vitro cardiovascular effects of PCBs. Oxidation of PCBs, especially to di-hydroxylated and quinone metabolites, was associated with the most pronounced effects, whereas sulfation and methoxylation of PCBs resulted in diminished bioactivity.

DISCUSSION

Risk characterization analysis showed that although in vitro derived effective concentrations exceeded the levels measured in the general population, risks cannot be ruled out due to the potential for population variability in susceptibility and the need to fill data gaps using read-across approaches. This study demonstrated a strategy for how in vitro data can be used to characterize human health risks from PCBs and their metabolites. https://doi.org/10.1289/EHP7030.

Rapid hazard characterization of environmental chemicals using a compendium of human cell lines from different organs

The lack of adequate toxicity data for the vast majority of chemicals in the environment has spurred the development of new approach methodologies (NAMs). This study aimed to develop a practical high-throughput in vitro model for rapidly evaluating potential hazards of chemicals using a small number of human cells. Forty-two compounds were tested using human induced pluripotent stem cell (iPSC)-derived cells (hepatocytes, neurons, cardiomyocytes and endothelial cells), and a primary endothelial cell line. Both functional and cytotoxicity endpoints were evaluated using high-content imaging. Concentration-response was used to derive points-of-departure (POD). PODs were integrated with ToxPi and used as surrogate NAM-based PODs for risk characterization. We found chemical class-specific similarity among the chemicals tested; metal salts exhibited the highest overall bioactivity. We also observed cell type-specific patterns among classes of chemicals, indicating the ability of the proposed in vitro model to recognize effects on different cell types. Compared to available NAM datasets, such as ToxCast/Tox21 and chemical structure-based descriptors, we found that the data from the five-cell-type model was as good or even better in assigning compounds to chemical classes. Additionally, the PODs from this model performed well as a conservative surrogate for regulatory in vivo PODs and were less likely to underestimate in vivo potency and potential risk compared to other NAM-based PODs. In summary, we demonstrate the potential of this in vitro screening model to inform rapid risk-based decision-making through ranking, clustering, and assessment of both hazard and risks of diverse environmental chemicals.

Transcriptomics in Toxicogenomics, Part III: Data Modelling for Risk Assessment

Transcriptomics data are relevant to address a number of challenges in Toxicogenomics (TGx). After careful planning of exposure conditions and data preprocessing, the TGx data can be used in predictive toxicology, where more advanced modelling techniques are applied. The large volume of molecular profiles produced by omics-based technologies allows the development and application of artificial intelligence (AI) methods in TGx. Indeed, the publicly available omics datasets are constantly increasing together with a plethora of different methods that are made available to facilitate their analysis, interpretation and the generation of accurate and stable predictive models. In this review, we present the state-of-the-art of data modelling applied to transcriptomics data in TGx. We show how the benchmark dose (BMD) analysis can be applied to TGx data. We review read across and adverse outcome pathways (AOP) modelling methodologies. We discuss how network-based approaches can be successfully employed to clarify the mechanism of action (MOA) or specific biomarkers of exposure. We also describe the main AI methodologies applied to TGx data to create predictive classification and regression models and we address current challenges. Finally, we present a short description of deep learning (DL) and data integration methodologies applied in these contexts. Modelling of TGx data represents a valuable tool for more accurate chemical safety assessment. This review is the third part of a three-article series on Transcriptomics in Toxicogenomics.

Toward Rigorous Materials Production: New Approach Methodologies Have Extensive Potential to Improve Current Safety Assessment Practices

Advanced material development, including at the nanoscale, comprises costly and complex challenges coupled to ensuring human and environmental safety. Governmental agencies regulating safety have announced interest toward acceptance of safety data generated under the collective term New Approach Methodologies (NAMs), as such technologies/approaches offer marked potential to progress the integration of safety testing measures during innovation from idea to product launch of nanomaterials. Divided in overall eight main categories, searchable databases for grouping and read across purposes, exposure assessment and modeling, in silico modeling of physicochemical structure and hazard data, in vitro high-throughput and high-content screening assays, dose-response assessments and modeling, analyses of biological processes and toxicity pathways, kinetics and dose extrapolation, consideration of relevant exposure levels and biomarker endpoints typify such useful NAMs. Their application generally agrees with articulated stakeholder needs for improvement of safety testing procedures. They further fit for inclusion and add value in nanomaterials risk assessment tools. Overall 37 of 50 evaluated NAMs and tiered workflows applying NAMs are recommended for considering safer-by-design innovation, including guidance to the selection of specific NAMs in the eight categories. An innovation funnel enriched with safety methods is ultimately proposed under the central aim of promoting rigorous nanomaterials innovation.

Transitioning the Generalised Read-Across approach (GenRA) to quantitative predictions: A case study using acute oral toxicity data

Read-across approaches continue to evolve as does their utility in the field of risk assessment. Previously we presented our generalised read-across (GenRA) approach (Shah et al., 2016), which utilises chemical descriptor and/or in vitro bioactivity data to make read-across predictions on the basis of the similarity weighted average of nearest neighbours. The current public version of GenRA predicts 574 apical outcomes as a binary call from repeat dose toxicity studies available in ToxRefDB (Helman et al., 2019). Here we investigated the application of GenRA to quantitative values, specifically using a large dataset of rat oral acute LD50 toxicity data (LD50 values for 7011 discrete chemicals) that had been collected under the auspices of the ICCVAM acute toxicity workgroup (ATWG). GenRA LD50 predictions were made based on the following criteria - chemicals were characterised by Morgan chemical fingerprints with a minimum similarity threshold of 0.5 and a maximum of 10 nearest neighbours over the entire dataset. An R² value of 0.61 and RMSE of 0.58 was achieved based on these parameters. Monte Carlo cross validation was then used to estimate confidence in the R². Cross validated R² values were found to fall in the range of 0.47 to 0.62. However, when evaluating GenRA locally to clusters of mechanistically or structurally-similar chemicals, average R² values improved up to 0.91. GenRA can be extended to make reasonable quantitative predictions of acute oral rodent toxicity with improved performance exhibited for specific local domains.

The Grouping and Assessment Strategy for Organic Pigments (GRAPE): Scientific evidence to facilitate regulatory decision-making

This article presents the Grouping and Assessment Strategy for Organic Pigments (GRAPE). GRAPE is driven by the hypotheses that low (bio)dissolution and low permeability indicate absence of systemic bioavailability and hence no systemic toxicity potential upon oral exposure, and, for inhalation exposure, that low (bio)dissolution (and absence of surface reactivity, dispersibility and in vitro effects) indicate that the organic pigment is a 'poorly soluble particle without intrinsic toxicity potential'. In GRAPE Tier 1, (bio)solubility and (bio)dissolution are assessed, and in Tier 2, in vitro Caco-2 permeability and in vitro alveolar macrophage activation. Thereafter, organic pigments are grouped by common properties (further considering structural similarity depending on the regulatory requirements). In Tier 3, absence of systemic bioavailability is verified by limited in vivo screening (rat 28-day oral and 5-day inhalation toxicity studies). If Tier 3 confirms no (or only very low) systemic bioavailability, all higher-tier endpoint-specific animal testing is scientifically not-relevant. Application of the GRAPE can serve to reduce animal testing needs for all but few representative organic pigments within a group. GRAPE stands in line with the EU REACH Regulation (Registration, Evaluation, Authorisation and Restriction of Chemicals). An ongoing research project aims at establishing a proof-of-concept of the GRAPE.

Quantitative prediction of repeat dose toxicity values using GenRA

Computational approaches have recently gained popularity in the field of read-across to automatically fill data-gaps for untested chemicals. Previously, we developed the generalized read-across (GenRA) tool, which utilizes in vitro bioactivity data in conjunction with chemical descriptor information to derive local validity domains to predict hazards observed in in vivo toxicity studies. Here, we modified GenRA to quantitatively predict point of departure (POD) values obtained from US EPA's Toxicity Reference Database (ToxRefDB) version 2.0. To evaluate GenRA predictions, we first aggregated oral Lowest Observed Adverse Effect Levels (LOAEL) for 1,014 chemicals by systemic, developmental, reproductive, and cholinesterase effects. The mean LOAEL values for each chemical were converted to log molar equivalents. Applying GenRA to all chemicals with a minimum Jaccard similarity threshold of 0.05 for Morgan fingerprints and a maximum of 10 nearest neighbors predicted systemic, developmental, reproductive, and cholinesterase inhibition min aggregated LOAEL values with R² values of 0.23, 0.22, 0.14, and 0.43, respectively. However, when evaluating GenRA locally to clusters of structurally-similar chemicals (containing 2 to 362 chemicals), average R² values for systemic, developmental, reproductive, and cholinesterase LOAEL predictions improved to 0.73, 0.66, 0.60 and 0.79, respectively. Our findings highlight the complexity of the chemical-toxicity landscape and the importance of identifying local domains where GenRA can be used most effectively for predicting PODs.

Using a hybrid read-across method to evaluate chemical toxicity based on chemical structure and biological data

Read-across has become a primary approach to fill data gaps for chemical safety assessments. Chemical similarity based on structure, reactivity, and physic-chemical property information is a traditional approach applied for read-across toxicity studies. However, toxicity mechanisms are usually complicated in a biological system, so only using chemical similarity to perform the read-across for new compounds was not satisfactory for most toxicity endpoints, especially when the chemically similar compounds show dissimilar toxicities. This study aims to develop an enhanced read-across method for chemical toxicity predictions. To this end, we used two large toxicity datasets for read-across purposes. One consists of 3979 compounds with Ames mutagenicity data, and the other contains 7332 compounds with rat acute oral toxicity data. First, biological data for all compounds in these two datasets were obtained by querying thousands of PubChem bioassays. The PubChem bioassays with at least five compounds from either of these two datasets showing active responses were selected to generate comprehensive bioprofiles. The read-across studies were performed by using chemical similarity search only and also by using a hybrid similarity search based on both chemical descriptors and bioprofiles. Compared to traditional read-across based on chemical similarity, the hybrid read-across approach showed improved accuracy of predictions for both Ames mutagenicity and acute oral toxicity. Furthermore, we could illustrate potential toxicity mechanisms by analyzing the bioprofiles used for this hybrid read-across study. The results of this study indicate that the new hybrid read-across approach could be an applicable computational tool for chemical toxicity predictions. In this way, the bottleneck of traditional read-across studies can be overcome by introducing public biological data into the traditional process. The incorporation of bioprofiles generated from the additional biological data for compounds can partially solve the "activity cliff" issue and reveal their potential toxicity mechanisms. This study leads to a promising direction to utilize data-driven approaches for computational toxicology studies in the big data era.

Exploring current read-across applications and needs among selected U.S. Federal Agencies

Read-across is a well-established data gap-filling technique applied for regulatory purposes. In US Environmental Protection Agency's New Chemicals Program under TSCA, read-across has been used extensively for decades, however the extent of application and acceptance of read-across among U.S. federal agencies is less clear. In an effort to build read-across capacity, raise awareness of the state of the science, and work towards a harmonization of read-across approaches across U.S. agencies, a new read-across workgroup was established under the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM). This is one of several ad hoc groups ICCVAM has convened to implement the ICCVAM Strategic Roadmap. In this article, we outline the charge and scope of the workgroup and summarize the current applications, tools used, and needs of the agencies represented on the workgroup for read-across. Of the agencies surveyed, the Environmental Protection Agency had the greatest experience in using read-across whereas other agencies indicated that they would benefit from gaining a perspective of the landscape of the tools and available guidance. Two practical case studies are also described to illustrate how the read-across approaches applied by two agencies vary on account of decision context.

Nonanimal Models for Acute Toxicity Evaluations: Applying Data-Driven Profiling and Read-Across

BACKGROUND

Low-cost, high-throughput in vitro bioassays have potential as alternatives to animal models for toxicity testing. However, incorporating in vitro bioassays into chemical toxicity evaluations such as read-across requires significant data curation and analysis based on knowledge of relevant toxicity mechanisms, lowering the enthusiasm of using the massive amount of unstructured public data.

OBJECTIVE

We aimed to develop a computational method to automatically extract useful bioassay data from a public repository (i.e., PubChem) and assess its ability to predict animal toxicity using a novel bioprofile-based read-across approach.

METHODS

A training database containing 7,385 compounds with diverse rat acute oral toxicity data was searched against PubChem to establish in vitro bioprofiles. Using a novel subspace clustering algorithm, bioassay groups that may inform on relevant toxicity mechanisms underlying acute oral toxicity were identified. These bioassays groups were used to predict animal acute oral toxicity using read-across through a cross-validation process. Finally, an external test set of over 600 new compounds was used to validate the resulting model predictivity.

RESULTS

Several bioassay clusters showed high predictivity for acute oral toxicity (positive prediction rates range from 62-100%) through cross-validation. After incorporating individual clusters into an ensemble model, chemical toxicants in the external test set were evaluated for putative acute toxicity (positive prediction rate equal to 76%). Additionally, chemical fragment -in vitro-in vivo relationships were identified to illustrate new animal toxicity mechanisms.

CONCLUSIONS

The in vitro bioassay data-driven profiling strategy developed in this study meets the urgent needs of computational toxicology in the current big data era and can be extended to develop predictive models for other complex toxicity end points. https://doi.org/10.1289/EHP3614.

Advancing Computational Toxicology in the Big Data Era by Artificial Intelligence: Data-Driven and Mechanism-Driven Modeling for Chemical Toxicity

In 2016, the Frank R. Lautenberg Chemical Safety for the 21st Century Act became the first US legislation to advance chemical safety evaluations by utilizing novel testing approaches that reduce the testing of vertebrate animals. Central to this mission is the advancement of computational toxicology and artificial intelligence approaches to implementing innovative testing methods. In the current big data era, the terms volume (amount of data), velocity (growth of data), and variety (the diversity of sources) have been used to characterize the currently available chemical, in vitro, and in vivo data for toxicity modeling purposes. Furthermore, as suggested by various scientists, the variability (internal consistency or lack thereof) of publicly available data pools, such as PubChem, also presents significant computational challenges. The development of novel artificial intelligence approaches based on public massive toxicity data is urgently needed to generate new predictive models for chemical toxicity evaluations and make the developed models applicable as alternatives for evaluating untested compounds. In this procedure, traditional approaches (e.g., QSAR) purely based on chemical structures have been replaced by newly designed data-driven and mechanism-driven modeling. The resulting models realize the concept of adverse outcome pathway (AOP), which can not only directly evaluate toxicity potentials of new compounds, but also illustrate relevant toxicity mechanisms. The recent advancement of computational toxicology in the big data era has paved the road to future toxicity testing, which will significantly impact on the public health.

Generalized Read-Across (GenRA): A workflow implemented into the EPA CompTox Chemicals Dashboard

Generalized Read-Across (GenRA) is a data driven approach which makes read-across predictions on the basis of a similarity weighted activity of source analogues (nearest neighbors). GenRA has been described in more detail in the literature (Shah et al., 2016; Helman et al., 2018). Here we present its implementation within the EPA's CompTox Chemicals Dashboard to provide public access to a GenRA module structured as a read-across workflow. GenRA assists researchers in identifying source analogues, evaluating their validity and making predictions of in vivo toxicity effects for a target substance. Predictions are presented as binary outcomes reflecting presence or absence of toxicity together with quantitative measures of uncertainty. The approach allows users to identify analogues in different ways, quickly assess the availability of relevant in vivo data for those analogues and visualize these in a data matrix to evaluate the consistency and concordance of the available experimental data for those analogues before making a GenRA prediction. Predictions can be exported into a tab-separated value (TSV) or Excel file for additional review and analysis (e.g., doses of analogues associated with production of toxic effects).  GenRA offers a new capability of making reproducible read-across predictions in an easy-to use-interface.

Multi-dimensional in vitro bioactivity profiling for grouping of glycol ethers

High-content screening data derived from physiologically-relevant in vitro models promise to improve confidence in data-integrative groupings for read-across in human health safety assessments. The biological data-based read-across concept is especially applicable to bioactive chemicals with defined mechanisms of toxicity; however, the challenge of data-derived groupings for chemicals that are associated with little or no bioactivity has not been explored. In this study, we apply a suite of organotypic and population-based in vitro models for comprehensive bioactivity profiling of twenty E-Series and P-Series glycol ethers, solvents with a broad variation in toxicity ranging from relatively non-toxic to reproductive and hematopoetic system toxicants. Both E-Series and P-Series glycol ethers elicited cytotoxicity only at high concentrations (mM range) in induced pluripotent stem cell-derived hepatocytes and cardiomyocytes. Population-variability assessment comprised a study of cytotoxicity in 94 human lymphoblast cell lines from 9 populations and revealed differences in inter-individual variability across glycol ethers, but did not indicate population-specific effects. Data derived from various phenotypic and transcriptomic assays revealed consistent bioactivity trends between both cardiomyocytes and hepatocytes, indicating a more universal, rather than cell-type specific mode-of-action for the tested glycol ethers in vitro. In vitro bioactivity-based similarity assessment using Toxicological Priority Index (ToxPi) showed that glycol ethers group according to their alcohol chain length, longer chains were associated with increased bioactivity. While overall in vitro bioactivity profiles did not correlate with in vivo toxicity data on glycol ethers, in vitro bioactivity of E-series glycol ethers were indicative of and correlated with in vivo irritation scores.