Formation of Structural Categories to Allow for Read-Across for Teratogenicity

Physiologically based kinetic (PBK) modeling of propiconazole using a machine learning-enhanced read-across approach for interspecies extrapolation

A significant challenge in the traditional human health risk assessment of agrochemicals is the uncertainty in quantifying the interspecies differences between animal models and humans. To work toward a more accurate and animal-free risk determination, new approaches such as physiologically based kinetic (PBK) modeling have been used to perform dosimetry extrapolation from animals to humans. However, the regulatory use and acceptance of PBK modeling is limited for chemicals that lack in vivo animal pharmacokinetic (PK) data, given the inability to evaluate models. To address these challenges, this study developed PBK models in the absence of in vivo PK data for the fungicide propiconazole, an activator of constitutive androstane receptor (CAR)/pregnane X receptor (PXR). A fit-for-purpose read-across approach was integrated with hierarchical clustering - an unsupervised machine learning algorithm, to bridge the knowledge gap. The integration allowed the incorporation of a broad spectrum of attributes for analog consideration, and enabled the analog selection in a simple, reproducible, and objective manner. The applicability was evaluated and demonstrated using penconazole (source) and three pseudo-unknown target chemicals (epoxiconazole, tebuconazole and triadimefon). Applying this machine learning-enhanced read-across approach, difenoconazole was selected as the most appropriate analog for propiconazole. A mouse PBK model was developed and evaluated for difenoconazole (source), with the mode of action of CAR/PXR activation incorporated to simulate the in vivo autoinduction of metabolism. The difenoconazole mouse model then served as a template for constructing the propiconazole mouse model. A parallelogram approach was subsequently applied to develop the propiconazole rat and human models, enabling a quantitative assessment of interspecies differences in dosimetry. This integrated approach represents a substantial advancement toward refining risk assessment of propiconazole within the framework of animal alternative safety assessment strategies.

Pyruvate dehydrogenase is a potential mitochondrial off-target for gentamicin based on in silico predictions and in vitro inhibition studies

During the drug development process, organ toxicity leads to an estimated failure of one-third of novel chemical entities. Drug-induced toxicity is increasingly associated with mitochondrial dysfunction, but identifying the underlying molecular mechanisms remains a challenge. Computational modeling techniques have proven to be a good tool in searching for drug off-targets. Here, we aimed to identify mitochondrial off-targets of the nephrotoxic drugs tenofovir and gentamicin using different in silico approaches (KRIPO, ProBis and PDID). Dihydroorotate dehydrogenase (DHODH) and pyruvate dehydrogenase (PDH) were predicted as potential novel off-target sites for tenofovir and gentamicin, respectively. The predicted targets were evaluated in vitro, using (colorimetric) enzymatic activity measurements. Tenofovir did not inhibit DHODH activity, while gentamicin potently reduced PDH activity. In conclusion, the use of in silico methods appeared a valuable approach in predicting PDH as a mitochondrial off-target of gentamicin. Further research is required to investigate the contribution of PDH inhibition to overall renal toxicity of gentamicin.

Sub-structure-based category formation for the prioritisation of genotoxicity hazard assessment for pesticide residues: Sulphonyl ureas

In dietary risk assessment, residues of pesticidal ingredients or their metabolites need to be evaluated for their genotoxic potential. The European Food Safety Authority recommend a tiered approach focussing assessment and testing on classes of similar chemicals. To characterise similarity and to identify structural alerts associated with genotoxic concern, a set of chemical sub-structures was derived for an example dataset of 74 sulphonyl urea agrochemicals for which either Ames, chromosomal aberration or micronucleus test results are publicly available. This analysis resulted in a set of seven structural alerts that define the chemical space, in terms of the common parent and metabolic scaffolds, associated with the sulphonyl urea chemical class. An analysis of the available profiling schemes for DNA and protein reactivity shows the importance of investigating the predictivity of such schemes within a well-defined area of structural space. Structural space alerts, covalent chemistry profiling and physico-chemistry properties were combined to develop chemical categories suitable for chemical prioritisation. The method is a robust and reproducible approach to such read-across predictions, with the potential to reduce unnecessary testing. The key challenge in the approach was identified as being the need for pesticide-class specific metabolism data as the basis for structural space alert development.

Assessment of the predictive capacity of a physiologically based kinetic model using a read-across approach

•

Potential regulatory application of PBK modelling information to assist read-across.

•

Presents workflow to read across PBK model information from data-rich to data-poor chemicals.

•

Describes appropriate analogue selection based on a set of specific criteria.

•

Uses estragole and safrole as source chemicals for a target chemical - methyleugenol.

•

Example of PBK model validation where in vivo kinetic data are lacking.

With current progress in science, there is growing interest in developing and applying Physiologically Based Kinetic (PBK) models in chemical risk assessment, as knowledge of internal exposure to chemicals is critical to understanding potential effects in vivo. In particular, a new generation of PBK models is being developed in which the model parameters are derived from in silico and in vitro methods. To increase the acceptance and use of these “Next Generation PBK models”, there is a need to demonstrate their validity. However, this is challenging in the case of data-poor chemicals that are lacking in kinetic data and for which predictive capacity cannot, therefore, be assessed. The aim of this work is to lay down the fundamental steps in using a read across framework to inform modellers and risk assessors on how to develop, or evaluate, PBK models for chemicals without in vivo kinetic data. The application of a PBK model that takes into account the absorption, distribution, metabolism and excretion characteristics of the chemical reduces the uncertainties in the biokinetics and biotransformation of the chemical of interest. A strategic flow-charting application, proposed herein, allows users to identify the minimum information to perform a read-across from a data-rich chemical to its data-poor analogue(s). The workflow analysis is illustrated by means of a real case study using the alkenylbenzene class of chemicals, showing the reliability and potential of this approach. It was demonstrated that a consistent quantitative relationship between model simulations could be achieved using models for estragole and safrole (source chemicals) when applied to methyleugenol (target chemical). When the PBK model code for the source chemicals was adapted to utilise input values relevant to the target chemical, simulation was consistent between the models. The resulting PBK model for methyleugenol was further evaluated by comparing the results to an existing, published model for methyleugenol, providing further evidence that the approach was successful. This can be considered as a “read-across” approach, enabling a valid PBK model to be derived to aid the assessment of a data poor chemical.

Computational approach for collection and prediction of molecular initiating events in developmental toxicity

Developmental toxicity is defined as the occurrence of adverse effects on the developing organism as a result from exposure to a toxic agent. These alterations can have long-term acute effects. Current in vitro models present important limitations and the evaluation of toxicity is not entirely objective. In silico methods have also shown limited success, in part due to complex and varied mechanisms of action that mediate developmental toxicity, which are sometimes poorly understood. In this article, we compiled a dataset of compounds with developmental toxicity categories and annotated mechanisms of action for both toxic and non-toxic compounds (DVTOX). With it, we selected a panel of protein targets that might be part of putative Molecular Initiating Events (MIEs) of Adverse Outcome Pathways of developmental toxicity. The validity of this list of candidate MIEs was studied through the evaluation of new drug-target relationships that include such proteins, but were not part of the original database. Finally, an orthology analysis of this protein panel was conducted to select an appropriate animal model to assess developmental toxicity. We tested our approach using the zebrafish embryo toxicity test, finding positive results.

Support Vector Machine model for hERG inhibitory activities based on the integrated hERG database using descriptor selection by NSGA-II

Assessing the hERG liability in the early stages of drug discovery programs is important. The recent increase of hERG-related information in public databases enabled various successful applications of machine learning techniques to predict hERG inhibition. However, most of these researches constructed the datasets from only one database, limiting the predictability and scope of the models. In this study, a hERG classification model was constructed using the largest dataset for hERG inhibition built by integrating multiple databases. The integrated dataset consisted of more than 291,000 structurally diverse compounds derived from ChEMBL, GOSTAR, PubChem, and hERGCentral. The prediction model was built by support vector machine (SVM) with descriptor selection based on Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to optimize the descriptor set for maximum prediction performance with the minimal number of descriptors. The SVM classification model using 72 selected descriptors and ECFP_4 structural fingerprints recorded kappa statistics of 0.733 and accuracy of 0.984 for the test set, substantially outperforming the prediction performance of the current commercial applications for hERG prediction. Finally, the applicability domain of the prediction model was assessed based on the molecular similarity between the training set and test set compounds.

A Mechanistic Framework for Integrating Chemical Structure and High-Throughput Screening Results to Improve Toxicity Predictions

Adverse Outcome Pathways (AOPs) establish a connection between a molecular initiating event (MIE) and an adverse outcome. Detailed understanding of the MIE provides the ideal data for determining chemical properties required to elicit the MIE. This study utilized high-throughput screening data from the ToxCast program, coupled with chemical structural information, to generate chemical clusters using three similarity methods pertaining to nine MIEs within an AOP network for hepatic steatosis. Three case studies demonstrate the utility of the mechanistic information held by the MIE for integrating biological and chemical data. Evaluation of the chemical clusters activating the glucocorticoid receptor identified activity differences in chemicals within a cluster. Comparison of the estrogen receptor results with previous work showed that bioactivity data and structural alerts can be combined to improve predictions in a customizable way where bioactivity data are limited. The aryl hydrocarbon receptor (AHR) highlighted that while structural data can be used to offset limited data for new screening efforts, not all ToxCast targets have sufficient data to define robust chemical clusters. In this context, an alternative to additional receptor assays is proposed where assays for proximal key events downstream of AHR activation could be used to enhance confidence in active calls. These case studies illustrate how the AOP framework can support an iterative process whereby in vitro toxicity testing and chemical structure can be combined to improve toxicity predictions. In vitro assays can inform the development of structural alerts linking chemical structure to toxicity. Consequently, structurally related chemical groups can facilitate identification of assays that would be informative for a specific MIE. Together, these activities form a virtuous cycle where the mechanistic basis for the in vitro results and the breadth of the structural alerts continually improve over time to better predict activity of chemicals for which limited toxicity data exist.

In Silico Prediction of Organ Level Toxicity: Linking Chemistry to Adverse Effects

In silico methods to predict toxicity include the use of (Quantitative) Structure-Activity Relationships ((Q)SARs) as well as grouping (category formation) allowing for read-across. A challenging area for in silico modelling is the prediction of chronic toxicity and the No Observed (Adverse) Effect Level (NO(A)EL) in particular. A proposed solution to the prediction of chronic toxicity is to consider organ level effects, as opposed to modelling the NO(A)EL itself. This review has focussed on the use of structural alerts to identify potential liver toxicants. In silico profilers, or groups of structural alerts, have been developed based on mechanisms of action and informed by current knowledge of Adverse Outcome Pathways. These profilers are robust and can be coded computationally to allow for prediction. However, they do not cover all mechanisms or modes of liver toxicity and recommendations for the improvement of these approaches are given.

Integrated Computational Solution for Predicting Skin Sensitization Potential of Molecules

Introduction

Skin sensitization forms a major toxicological endpoint for dermatology and cosmetic products. Recent ban on animal testing for cosmetics demands for alternative methods. We developed an integrated computational solution (SkinSense) that offers a robust solution and addresses the limitations of existing computational tools i.e. high false positive rate and/or limited coverage.

Results

The key components of our solution include: QSAR models selected from a combinatorial set, similarity information and literature-derived sub-structure patterns of known skin protein reactive groups. Its prediction performance on a challenge set of molecules showed accuracy = 75.32%, CCR = 74.36%, sensitivity = 70.00% and specificity = 78.72%, which is better than several existing tools including VEGA (accuracy = 45.00% and CCR = 54.17% with ‘High’ reliability scoring), DEREK (accuracy = 72.73% and CCR = 71.44%) and TOPKAT (accuracy = 60.00% and CCR = 61.67%). Although, TIMES-SS showed higher predictive power (accuracy = 90.00% and CCR = 92.86%), the coverage was very low (only 10 out of 77 molecules were predicted reliably).

Conclusions

Owing to improved prediction performance and coverage, our solution can serve as a useful expert system towards Integrated Approaches to Testing and Assessment for skin sensitization. It would be invaluable to cosmetic/ dermatology industry for pre-screening their molecules, and reducing time, cost and animal testing.

Toward a unifying strategy for the structure-based prediction of toxicological endpoints

Most computational methods used for the prediction of toxicity endpoints are based on the assumption that similar compounds have similar biological properties. This principle can be exploited using computational methods like read across or quantitative structure-activity relationships. However, there is no general agreement about which method is the most appropriate for quantifying compound similarity neither for exploiting the similarity principle in order to obtain reliable estimations of the compound properties. Moreover, optimal similarity metrics and modeling methods might depend on the characteristics of the endpoints and training series used in each case. This study describes a comparative analysis of the predictive performance of diverse similarity metrics and modeling methods in toxicological applications. A collection of two quantitative (n = 660, n = 1114) and three qualitative (n = 447, n = 905, n = 1220) datasets representing very different endpoints of interest in drug safety evaluation and rigorous methods were used to estimate the external predictive ability in each case. The results confirm that no single approach produces the best results in all instances, and the best predictions were obtained using different tools in different situations. The trends observed in this study were exploited to propose a unifying strategy allowing the use of the most suitable method for every compound. A comparison of the quality of the predictions obtained by the unifying strategy with those obtained by standard prediction methods confirmed the usefulness of the proposed approach.

Novel approach for efficient predictions properties of large pool of nanomaterials based on limited set of species: nano-read-across

Creating suitable chemical categories and developing read-across methods, supported by quantum mechanical calculations, can be an effective solution to solving key problems related to current scarcity of data on the toxicity of various nanoparticles. This study has demonstrated that by applying a nano-read-across, the cytotoxicity of nano-sized metal oxides could be estimated with a similar level of accuracy as provided by quantitative structure-activity relationship for nanomaterials (nano-QSAR model(s)). The method presented is a suitable computational tool for the preliminary hazard assessment of nanomaterials. It also could be used for the identification of nanomaterials that may pose potential negative impact to human health and the environment. Such approaches are especially necessary when there is paucity of relevant and reliable data points to develop and validate nano-QSAR models.

Approaches for Grouping Chemicals into Categories

This chapter outlines the various methods by which chemical similarity can be defined to allow for chemical category formation. The focus is on three methods: simple analogues, definition of the chemistry associated with molecular initiating events and chemoinformatics. An outline of how each method is used in practice and how they have been developed into in silico tools is presented.

Alternative (non-animal) methods for cosmetics testing: current status and future prospects-2010

The 7th amendment to the EU Cosmetics Directive prohibits to put animal-tested cosmetics on the market in Europe after 2013. In that context, the European Commission invited stakeholder bodies (industry, non-governmental organisations, EU Member States, and the Commission's Scientific Committee on Consumer Safety) to identify scientific experts in five toxicological areas, i.e. toxicokinetics, repeated dose toxicity, carcinogenicity, skin sensitisation, and reproductive toxicity for which the Directive foresees that the 2013 deadline could be further extended in case alternative and validated methods would not be available in time. The selected experts were asked to analyse the status and prospects of alternative methods and to provide a scientifically sound estimate of the time necessary to achieve full replacement of animal testing. In summary, the experts confirmed that it will take at least another 7-9 years for the replacement of the current in vivo animal tests used for the safety assessment of cosmetic ingredients for skin sensitisation. However, the experts were also of the opinion that alternative methods may be able to give hazard information, i.e. to differentiate between sensitisers and non-sensitisers, ahead of 2017. This would, however, not provide the complete picture of what is a safe exposure because the relative potency of a sensitiser would not be known. For toxicokinetics, the timeframe was 5-7 years to develop the models still lacking to predict lung absorption and renal/biliary excretion, and even longer to integrate the methods to fully replace the animal toxicokinetic models. For the systemic toxicological endpoints of repeated dose toxicity, carcinogenicity and reproductive toxicity, the time horizon for full replacement could not be estimated.

The Use of a Chemistry-based Profiler for Covalent DNA Binding in the Development of Chemical Categories for Read-across for Genotoxicity

An important molecular initiating event for genotoxicity is the ability of a compound to bind covalently with DNA. However, not all compounds that can undergo covalent binding mechanisms will result in genotoxicity. One approach to solving this problem, when in silico prediction techniques are being used, is to develop tools that allow chemicals to be grouped into categories based on their ability to bind covalently to DNA. For this analysis to take place, compounds need to be placed within categories where the trend in toxicity can be explained by simple descriptors, such as hydrophobicity. However, this can occur only when the compounds within a category are structurally and mechanistically similar. Chemistry-based profilers have the ability to screen compounds and highlight those with similar structures to a target compound, and are thus likely to act via a similar mechanism of action. Here, examples are reported to highlight how structure-based profilers can be used to form categories and hence fill data gaps. The importance of developing a well-defined and robust category is discussed in terms of both mechanisms of action and structural similarity.

A review of the electrophilic reaction chemistry involved in covalent DNA binding

The need to assess the ability of a chemical to act as a mutagen or a genotoxic carcinogen (collectively termed genotoxicity) is one of the primary requirements in regulatory toxicology. Several pieces of legislation have led to an increased interest in the use of in silico methods, specifically the formation of chemical categories for the assessment of toxicological endpoints. A key step in the development of chemical categories for genotoxicity is defining the organic chemistry associated with the formation of a covalent bond between DNA and an exogenous chemical. This organic chemistry is typically defined as structural alerts. To this end, this article has reviewed the literature defining the structural alerts associated with covalent DNA binding. Importantly, this review article also details the mechanistic organic chemistry associated with each of the structural alerts. This information is extremely important in terms of meeting regulatory requirements for the acceptance of the chemical category approach. The structural alerts and associated mechanistic chemistry have been incorporated into the Organisation for Economic Co-operation and Development (OECD) (Q)SAR Application Toolbox.

Open Source Tools for Read-Across and Category Formation

In this chapter, the requirements and challenges for computational tools to support category formation and read-across are summarised. A brief overview of the open source, open data and open standards approaches in chemoinformatics are presented. The opportunities offered by these kinds of “openness” are highlighted, with emphasis on open source applications specifically developed to address challenges posed by the REACH regulation. Structural similarity assessment is currently a common practice in forming categories and applying read-across, and in developing and validating (Q)SARs. The Toxmatch software provides several endpoint-specific similarity measures, with descriptors selected using a training set in combination with data mining methods. The Toxtree software implements several classification schemes for predicting various endpoints and relies primarily on chemical structure, metabolic pathways, physico-chemical properties and descriptors, calculated from chemical structure. In addition to making predictions for individual chemicals, Toxtree can be used to profile the toxicological hazard or mechanistic group of a set of chemicals. Ambit is anopen source software for chemoinformatics data management, which allows storage of a large number of chemical structures and toxicological data and provides a flexible means for exploration of structural and similarity spaces. Several examples of the application of read-across, initiated by an expert-defined search strategy and supported by Ambit search functionalities are described.

The use of Mechanisms and Modes of Toxic Action in Integrated Testing Strategies: The Report and Recommendations of a Workshop held as part of the European Union OSIRIS Integrated Project

This report on The Potential of Mode of Action (MoA) Information Derived from Non-testing and Screening Methodologies to Support Informed Hazard Assessment, resulted from a workshop organised within OSIRIS (Optimised Strategies for Risk Assessment of Industrial Chemicals through Integration of Non-test and Test Information), a project partly funded by the EU Commission within the Sixth Framework Programme. The workshop was held in Liverpool, UK, on 30 October 2008, with 35 attendees. The goal of the OSIRIS project is to develop integrated testing strategies (ITS) fit for use in the REACH system, that would enable a significant increase in the use of non-testing information for regulatory decision making, and thus minimise the need for animal testing. One way to improve the evaluation of chemicals may be through categorisation by way of mechanisms or modes of toxic action. Defining such groups can enhance read-across possibilities and priority settings for certain toxic modes or chemical structures responsible for these toxic modes. Overall, this may result in a reduction of in vivo testing on organisms, through combining available data on mode of action and a focus on the potentially most-toxic groups. In this report, the possibilities of a mechanistic approach to assist in and guide ITS are explored, and the differences between human health and environmental areas are summarised.

Substance-tailored testing strategies in toxicology: an in silico methodology based on QSAR modeling of toxicological thresholds and Monte Carlo simulations of toxicological testing

The design of toxicological testing strategies aimed at identifying the toxic effects of chemicals without (or with a minimal) recourse to animal experimentation is an important issue for toxicological regulations and for industrial decision-making. This article describes an original approach which enables the design of substance-tailored testing strategies with a specified performance in terms of false-positive and false-negative rates. The outcome of toxicological testing is simulated in a different way than previously published articles on the topic. Indeed, toxicological outcomes are simulated not only as a function of the performance of toxicological tests but also as a function of the physico-chemical properties of chemicals. The required inputs for our approach are QSAR predictions for the LOAELs of the toxicological effect of interest and statistical distributions describing the relationship existing between in vivo LOAEL values and results from in vitro tests. Our methodology is able to correctly predict the performance of testing strategies designed to analyze the teratogenic effects of two chemicals: di(2-ethylhexyl)phthalate and Indomethacin. The proposed decision-support methodology can be adapted to any toxicological context as long as a statistical comparison between in vitro and in vivo results is possible and QSAR models for the toxicological effect of interest can be developed.