High Throughput Read-Across for Screening a Large Inventory of Related Structures by Balancing Artificial Intelligence/Machine Learning and Human Knowledge.

Read-across is an in silico method applied in chemical risk assessment for data-poor chemicals. The read-across outcomes for repeated-dose toxicity end points include the no-observed-adverse-effect level (NOAEL) and estimated uncertainty for a particular category of effects. We have previously developed a new paradigm for estimating NOAELs based on chemoinformatics analysis and experimental study qualities from selected analogues, not relying on quantitative structure-activity relationships (QSARs) or rule-based SAR systems, which are not well-suited to end points for which the underpinning data are weakly grounded in specific chemical-biological interactions. The central hypothesis of this approach is that similar compounds have similar toxicity profiles and, hence, similar NOAEL values. Analogue quality (AQ) quantifies the suitability of an analogue candidate for reading across to the target by considering similarity from structure, physicochemical, ADME (absorption, distribution, metabolism, excretion), and biological perspectives. Biological similarity is based on experimental data; assay vectors derived from aggregations of ToxCast/Tox21 data are used to derive machine learning (ML) hybrid rules that serve as biological fingerprints to capture target-analogue similarity relevant to specific effects of interest, for example, hormone receptors (ER/AR/THR). Once one or more analogues have been qualified for read-across, a decision theory approach is used to estimate confidence bounds for the NOAEL of the target. The confidence interval is dramatically narrowed when analogues are constrained to biologically related profiles. Although this read-across process works well for a single target with several analogues, it can become unmanageable when, for example, screening multiple targets (e.g., virtual screening library) or handling a parent compound having numerous metabolites. To this end, we have established a digitalized framework to enable the assessment of a large number of substances, while still allowing for human decisions for filtering and prioritization. This workflow was developed and validated through a use case of a large set of bisphenols and their metabolites.

Phthalates and alternative plasticizers differentially affect phenotypic parameters in gonadal somatic and germ cell lines†.

The developmental and reproductive toxicity associated with exposure to phthalates has motivated a search for alternatives. However, there is limited knowledge regarding the adverse effects of some of these chemicals. We used high-content imaging to compare the effects of mono (2-ethylhexyl) phthalate (MEHP) with six alternative plasticizers: di-2-ethylhexyl terephthalate (DEHTP); diisononyl-phthalate (DINP); di-isononylcyclohexane-1,2-dicarboxylate (DINCH); 2-ethylhexyl adipate (DEHA); 2,2,4-trimethyl 1,3-pentanediol diisobutyrate (TXIB) and di-iso-decyl-adipate (DIDA). A male germ spermatogonial cell line (C18-4), a Sertoli cell line (TM4) and two steroidogenic cell lines (MA-10 Leydig and KGN granulosa) were exposed for 48 h to each chemical (0.001-100 µM). Cell images were analyzed to assess cytotoxicity and effects on phenotypic endpoints. Only MEHP (100 µM) was cytotoxic and only in C18-4 cells. However, several plasticizers had distinct phenotypic effects in all four cell lines. DINP increased Calcein intensity in C18-4 cells, whereas DIDA induced oxidative stress. In TM4 cells, MEHP, and DINCH affected lipid droplet numbers, while DEHTP and DINCH increased oxidative stress. In MA-10 cells, MEHP increased lipid droplet areas and oxidative stress; DINP decreased the number of lysosomes, while DINP, DEHA, and DIDA altered mitochondrial activity. In KGN cells, MEHP, DINP and DINCH increased the number of lipid droplets, whereas DINP decreased the number of lysosomes, increased oxidative stress and affected mitochondria. The Toxicological Priority Index (ToxPi) provided a visual illustration of the cell line specificity of the effects on phenotypic parameters. The lowest administered equivalent doses were observed for MEHP. We propose that this approach may assist in screening alternative plasticizers.

Comprehensive interpretation of in vitro micronucleus test results for 292 chemicals: from hazard identification to risk assessment application.

Risk assessments are increasingly reliant on information from in vitro assays. The in vitro micronucleus test (MNvit) is a genotoxicity test that detects chromosomal abnormalities, including chromosome breakage (clastogenicity) and/or whole chromosome loss (aneugenicity). In this study, MNvit datasets for 292 chemicals, generated by the US EPA's ToxCast program, were evaluated using a decision tree-based pipeline for hazard identification. Chemicals were tested with 19 concentrations (n = 1) up to 200 µM, in the presence and absence of Aroclor 1254-induced rat liver S9. To identify clastogenic chemicals, %MN values at each concentration were compared to a distribution of batch-specific solvent controls; this was followed by cytotoxicity assessment and benchmark concentration (BMC) analyses. The approach classified 157 substances as positives, 25 as negatives, and 110 as inconclusive. Using the approach described in Bryce et al. (Environ Mol Mutagen 52:280-286, 2011), we identified 15 (5%) aneugens. IVIVE (in vitro to in vivo extrapolation) was employed to convert BMCs into administered equivalent doses (AEDs). Where possible, AEDs were compared to points of departure (PODs) for traditional genotoxicity endpoints; AEDs were generally lower than PODs based on in vivo endpoints. To facilitate interpretation of in vitro MN assay concentration-response data for risk assessment, exposure estimates were utilized to calculate bioactivity exposure ratio (BER) values. BERs for 50 clastogens and two aneugens had AEDs that approached exposure estimates (i.e., BER < 100); these chemicals might be considered priorities for additional testing. This work provides a framework for the use of high-throughput in vitro genotoxicity testing for priority setting and chemical risk assessment.

Do Similar Structures Have Similar No Observed Adverse Effect Level (NOAEL) Values? Exploring Chemoinformatics Approaches for Estimating NOAEL Bounds and Uncertainties.

Determination of the no observed adverse effect level (NOAEL) of a substance is an important step in safety and regulatory assessments. Application of conventional in silico strategies, for example, quantitative structure-activity relationship (QSAR) models, to predict NOAEL values is inherently problematic. Whereas QSAR models for well-defined toxicity endpoints such as Ames mutagenicity or skin sensitization can be developed from mechanistic knowledge of molecular initiating events and adverse outcome pathways, QSAR is not appropriate for predicting a NOAEL value, a concentration at which "no effect" is observed. This paper presents a chemoinformatics approach and explores how it can be further refined through the incorporation of toxicity endpoint-specific information to estimate confidence bounds for the NOAEL of a target substance, given experimentally determined NOAEL values for one or more suitable analogues. With a sufficiently large NOAEL database, we analyze how a difference in NOAEL values for pairs of structures depends on their pairwise similarity, where similarity takes both structural features and physicochemical properties into account. The width of the estimate NOAEL confidence interval is proportional to the uncertainty. Using the new threshold of toxicological concern (TTC) database enriched with antimicrobials, examples are presented to illustrate how uncertainty decreases with increasing analogue quality and also how NOAEL bounds estimation can be significantly improved by filtering the full database to include only substances that are in structure categories relevant to the target and analogue.

Predicting estrogen receptor activation by a group of substituted phenols: An integrated approach to testing and assessment case study.

Traditional approaches for chemical risk assessment cannot keep pace with the number of substances requiring assessment. Thus, in a global effort to expedite and modernize chemical risk assessment, New Approach Methodologies (NAMs) are being explored and developed. Included in this effort is the OECD Integrated Approaches for Testing and Assessment (IATA) program, which provides a forum for OECD member countries to develop and present case studies illustrating the application of NAM in various risk assessment contexts. Here, we present an IATA case study for the prediction of estrogenic potential of three target phenols: 4-tert-butylphenol, 2,4-di-tert-butylphenol and octabenzone. Key features of this IATA include the use of two computational approaches for analogue selection for read-across, data collected from traditional and NAM sources, and a workflow to generate predictions regarding the targets' ability to bind the estrogen receptor (ER). Endocrine disruption can occur when a chemical substance mimics the activity of natural estrogen by binding to the ER and, if potency and exposure are sufficient, alters the function of the endocrine system to cause adverse effects. The data indicated that of the three target substances that were considered herein, 4-tert-butylphenol is a potential endocrine disruptor. Further, this IATA illustrates that the NAM approach explored is health protective when compared to in vivo endpoints traditionally used for human health risk assessment.

Genetic toxicology in silico protocol.

In silico toxicology (IST) approaches to rapidly assess chemical hazard, and usage of such methods is increasing in all applications but especially for regulatory submissions, such as for assessing chemicals under REACH as well as the ICH M7 guideline for drug impurities. There are a number of obstacles to performing an IST assessment, including uncertainty in how such an assessment and associated expert review should be performed or what is fit for purpose, as well as a lack of confidence that the results will be accepted by colleagues, collaborators and regulatory authorities. To address this, a project to develop a series of IST protocols for different hazard endpoints has been initiated and this paper describes the genetic toxicity in silico (GIST) protocol. The protocol outlines a hazard assessment framework including key effects/mechanisms and their relationships to endpoints such as gene mutation and clastogenicity. IST models and data are reviewed that support the assessment of these effects/mechanisms along with defined approaches for combining the information and evaluating the confidence in the assessment. This protocol has been developed through a consortium of toxicologists, computational scientists, and regulatory scientists across several industries to support the implementation and acceptance of in silico approaches.

Accelerating the Pace of Chemical Risk Assessment.

Changes in chemical regulations worldwide have increased the demand for new data on chemical safety. New approach methodologies (NAMs) are defined broadly here as including in silico approaches and in chemico and in vitro assays, as well as the inclusion of information from the exposure of chemicals in the context of hazard [European Chemicals Agency, " New Approach Methodologies in Regulatory Science ", 2016]. NAMs for toxicity testing, including alternatives to animal testing approaches, have shown promise to provide a large amount of data to fill information gaps in both hazard and exposure. In order to increase experience with the new data and to advance the applications of NAM data to evaluate the safety of data-poor chemicals, demonstration case studies have to be developed to build confidence in their usability. Case studies can be used to explore the domains of applicability of the NAM data and identify areas that would benefit from further research, development, and application. To ensure that this science evolves with direct input from and engagement by risk managers and regulatory decision makers, a workshop was convened among senior leaders from international regulatory agencies to identify common barriers for using NAMs and to propose next steps to address them. Central to the workshop were a series of collaborative case studies designed to explore areas where the benefits of NAM data could be demonstrated. These included use of in vitro bioassays data in combination with exposure estimates to derive a quantitative assessment of risk, use of NAMs for updating chemical categorizations, and use of NAMs to increase understanding of exposure and human health toxicity of various chemicals. The case study approach proved effective in building collaborations and engagement with regulatory decision makers and to promote the importance of data and knowledge sharing among international regulatory agencies. The case studies will be continued to explore new ways of describing hazard (i.e., pathway perturbations as a measure of adversity) and new ways of describing risk (i.e., using NAMs to identify protective levels without necessarily being predictive of a specific hazard). Importantly, the case studies also highlighted the need for increased training and communication across the various communities including the risk assessors, regulators, stakeholders (e.g., industry, non-governmental organizations), and the general public. The development and application of NAMs will play an increasing role in filling important data gaps on the safety of chemicals, but confidence in NAMs will only come with learning by doing and sharing in the experience.

Emerging technologies for food and drug safety.

Emerging technologies are playing a major role in the generation of new approaches to assess the safety of both foods and drugs. However, the integration of emerging technologies in the regulatory decision-making process requires rigorous assessment and consensus amongst international partners and research communities. To that end, the Global Coalition for Regulatory Science Research (GCRSR) in partnership with the Brazilian Health Surveillance Agency (ANVISA) hosted the seventh Global Summit on Regulatory Science (GSRS17) in Brasilia, Brazil on September 18-20, 2017 to discuss the role of new approaches in regulatory science with a specific emphasis on applications in food and medical product safety. The global regulatory landscape concerning the application of new technologies was assessed in several countries worldwide. Challenges and issues were discussed in the context of developing an international consensus for objective criteria in the development, application and review of emerging technologies. The need for advanced approaches to allow for faster, less expensive and more predictive methodologies was elaborated. In addition, the strengths and weaknesses of each new approach was discussed. And finally, the need for standards and reproducible approaches was reviewed to enhance the application of the emerging technologies to improve food and drug safety. The overarching goal of GSRS17 was to provide a venue where regulators and researchers meet to develop collaborations addressing the most pressing scientific challenges and facilitate the adoption of novel technical innovations to advance the field of regulatory science.

Recommended approaches in the application of toxicogenomics to derive points of departure for chemical risk assessment.

There is increasing interest in the use of quantitative transcriptomic data to determine benchmark dose (BMD) and estimate a point of departure (POD) for human health risk assessment. Although studies have shown that transcriptional PODs correlate with those derived from apical endpoint changes, there is no consensus on the process used to derive a transcriptional POD. Specifically, the subsets of informative genes that produce BMDs that best approximate the doses at which adverse apical effects occur have not been defined. To determine the best way to select predictive groups of genes, we used published microarray data from dose-response studies on six chemicals in rats exposed orally for 5, 14, 28, and 90 days. We evaluated eight approaches for selecting genes for POD derivation and three previously proposed approaches (the lowest pathway BMD, and the mean and median BMD of all genes). The relationship between transcriptional BMDs derived using these 11 approaches and PODs derived from apical data that might be used in chemical risk assessment was examined. Transcriptional BMD values for all 11 approaches were remarkably aligned with corresponding apical PODs, with the vast majority of toxicogenomics PODs being within tenfold of those derived from apical endpoints. We identified at least four approaches that produce BMDs that are effective estimates of apical PODs across multiple sampling time points. Our results support that a variety of approaches can be used to derive reproducible transcriptional PODs that are consistent with PODs produced from traditional methods for chemical risk assessment.

Increasing Scientific Confidence in Adverse Outcome Pathways: Application of Tailored Bradford-Hill Considerations for Evaluating Weight of Evidence.

Systematic consideration of scientific support is a critical element in developing and, ultimately, using adverse outcome pathways (AOPs) for various regulatory applications. Though weight of evidence (WoE) analysis has been proposed as a basis for assessment of the maturity and level of confidence in an AOP, methodologies and tools are still being formalized. The Organization for Economic Co-operation and Development (OECD) Users' Handbook Supplement to the Guidance Document for Developing and Assessing AOPs (OECD 2014a; hereafter referred to as the OECD AOP Handbook) provides tailored Bradford-Hill (BH) considerations for systematic assessment of confidence in a given AOP. These considerations include (1) biological plausibility and (2) empirical support (dose-response, temporality, and incidence) for Key Event Relationships (KERs), and (3) essentiality of key events (KEs). Here, we test the application of these tailored BH considerations and the guidance outlined in the OECD AOP Handbook using a number of case examples to increase experience in more transparently documenting rationales for assigned levels of confidence to KEs and KERs, and to promote consistency in evaluation within and across AOPs. The major lessons learned from experience are documented, and taken together with the case examples, should contribute to better common understanding of the nature and form of documentation required to increase confidence in the application of AOPs for specific uses. Based on the tailored BH considerations and defining questions, a prototype quantitative model for assessing the WoE of an AOP using tools of multi-criteria decision analysis (MCDA) is described. The applicability of the approach is also demonstrated using the case example aromatase inhibition leading to reproductive dysfunction in fish. Following the acquisition of additional experience in the development and assessment of AOPs, further refinement of parameterization of the model through expert elicitation is recommended. Overall, the application of quantitative WoE approaches hold promise to enhance the rigor, transparency and reproducibility for AOP WoE determinations and may play an important role in delineating areas where research would have the greatest impact on improving the overall confidence in the AOP.

Performance of (Q)SAR models for predicting Ames mutagenicity of aryl azo and benzidine based compounds.

Regulatory agencies worldwide are committed to the objectives of the Strategic Approach to International Chemicals Management to ensure that by 2020 chemicals are used and produced in ways that lead to the minimization of significant adverse effects on human health and the environment. Under the Government of Canada's Chemicals Management Plan, the commitment to address a large number of substances, many with limited data, has highlighted the importance of pursuing alternative hazard assessment methodologies that are able to accommodate chemicals with varying toxicological information. One such method is (Quantitative) Structure Activity Relationships ((Q)SAR) models. The current investigation into the predictivity of 20 (Q)SAR tools designed to model bacterial reverse mutation in Salmonella typhimurium is one of the first of this magnitude to be carried out using an external validation set comprised mainly of industrial chemicals which represent a diverse group of aromatic and benzidine-based azo dyes and pigments. Overall, this study highlights the value in challenging the predictivity of existing models using a small but representative subset of data-rich chemicals. Furthermore, external validation revealed that only a handful of models satisfactorily predicted for the test chemical space. The exercise also provides insight into using the Organisation for Economic Co-operation and Development (Q)SAR Toolbox as a read across tool.

Development and application of consensus in silico models for advancing high-throughput toxicological predictions.

Computational toxicology models have been successfully implemented to prioritize and screen chemicals. There are numerous in silico (quantitative) structure-activity relationship ([Q]SAR) models for the prediction of a range of human-relevant toxicological endpoints, but for a given endpoint and chemical, not all predictions are identical due to differences in their training sets, algorithms, and methodology. This poses an issue for high-throughput screening of a large chemical inventory as it necessitates several models to cover diverse chemistries but will then generate data conflicts. To address this challenge, we developed a consensus modeling strategy to combine predictions obtained from different existing in silico (Q)SAR models into a single predictive value while also expanding chemical space coverage. This study developed consensus models for nine toxicological endpoints relating to estrogen receptor (ER) and androgen receptor (AR) interactions (i.e., binding, agonism, and antagonism) and genotoxicity (i.e., bacterial mutation, in vitro chromosomal aberration, and in vivo micronucleus). Consensus models were created by combining different (Q)SAR models using various weighting schemes. As a multi-objective optimization problem, there is no single best consensus model, and therefore, Pareto fronts were determined for each endpoint to identify the consensus models that optimize the multiple-criterion decisions simultaneously. Accordingly, this work presents sets of solutions for each endpoint that contain the optimal combination, regardless of the trade-off, with the results demonstrating that the consensus models improved both the predictive power and chemical space coverage. These solutions were further analyzed to find trends between the best consensus models and their components. Here, we demonstrate the development of a flexible and adaptable approach for in silico consensus modeling and its application across nine toxicological endpoints related to ER activity, AR activity, and genotoxicity. These consensus models are developed to be integrated into a larger multi-tier NAM-based framework to prioritize chemicals for further investigation and support the transition to a non-animal approach to risk assessment in Canada.

Establishing a quantitative framework for regulatory interpretation of genetic toxicity dose-response data: Margin of exposure case study of 48 compounds with both in vivo mutagenicity and carcinogenicity dose-response data.

Quantitative relationships between carcinogenic potency and mutagenic potency have been previously examined using a benchmark dose (BMD)-based approach. We extended those analyses by using human exposure data for 48 compounds to calculate carcinogenicity-derived and genotoxicity-derived margin of exposure values (MOEs) that can be used to prioritize substances for risk management. MOEs for 16 of the 48 compounds were below 10,000, and consequently highlighted for regulatory concern. Of these, 15 were highlighted using genotoxicity-derived (micronucleus [MN] dose-response data) MOEs. A total of 13 compounds were highlighted using carcinogenicity-derived MOEs; 12 compounds were overlapping. MOEs were also calculated using transgenic rodent (TGR) mutagenicity data. For 10 of the 12 compounds examined using TGR data, the results similarly revealed that mutagenicity-derived MOEs yield regulatory decisions that correspond with those based on carcinogenicity-derived MOEs. The effect of benchmark response (BMR) on MOE determination was also examined. Reinterpretation of the analyses using a BMR of 50% indicated that four out of 15 compounds prioritized using MN-derived MOEs based on a default BMR of 5% would have been missed. The results indicate that regulatory decisions based on in vivo genotoxicity dose-response data would be consistent with those based on carcinogenicity dose-response data; in some cases, genotoxicity-based decisions would be more conservative. Going forward, and in the absence of carcinogenicity data, in vivo genotoxicity assays (MN and TGR) can be used to effectively prioritize substances for regulatory action. Routine use of the MOE approach necessitates the availability of reliable human exposure estimates, and consensus regarding appropriate BMRs for genotoxicity endpoints.

In vitro transcriptomic analyses reveal pathway perturbations, estrogenic activities, and potencies of data-poor BPA alternative chemicals.

Since initial regulatory action in 2010 in Canada, bisphenol A (BPA) has been progressively replaced by structurally related alternative chemicals. Unfortunately, many of these chemicals are data-poor, limiting toxicological risk assessment. We used high-throughput transcriptomics to evaluate potential hazards and compare potencies of BPA and 15 BPA alternative chemicals in cultured breast cancer cells. MCF-7 cells were exposed to BPA and 15 alternative chemicals (0.0005-100 µM) for 48 h. TempO-Seq (BioSpyder Inc) was used to examine global transcriptomic changes and estrogen receptor alpha (ERα)-associated transcriptional changes. Benchmark concentration (BMC) analysis was conducted to identify 2 global transcriptomic points of departure: (1) the lowest pathway median gene BMC and (2) the 25th lowest rank-ordered gene BMC. ERα activation was evaluated using a published transcriptomic biomarker and an ERα-specific transcriptomic point of departure was derived. Genes fitting BMC models were subjected to upstream regulator and canonical pathway analysis in Ingenuity Pathway Analysis. Biomarker analysis identified BPA and 8 alternative chemicals as ERα active. Global and ERα transcriptomic points of departure produced highly similar potency rankings with bisphenol AF as the most potent chemical tested, followed by BPA and bisphenol C. Further, BPA and transcriptionally active alternative chemicals enriched similar gene sets associated with increased cell division and cancer-related processes. These data provide support for future read-across applications of transcriptomic profiling for risk assessment of data-poor chemicals and suggest that several BPA alternative chemicals may cause hazards at similar concentrations to BPA.

Quantitative in vitro to in vivo extrapolation of genotoxicity data provides protective estimates of in vivo dose.

Genotoxicity assessment is a critical component in the development and evaluation of chemicals. Traditional genotoxicity assays (i.e., mutagenicity, clastogenicity, and aneugenicity) have been limited to dichotomous hazard classification, while other toxicity endpoints are assessed through quantitative determination of points-of-departures (PODs) for setting exposure limits. The more recent higher-throughput in vitro genotoxicity assays, many of which also provide mechanistic information, offer a powerful approach for determining defined PODs for potency ranking and risk assessment. In order to obtain relevant human dose context from the in vitro assays, in vitro to in vivo extrapolation (IVIVE) models are required to determine what dose would elicit a concentration in the body demonstrated to be genotoxic using in vitro assays. Previous work has demonstrated that application of IVIVE models to in vitro bioactivity data can provide PODs that are protective of human health, but there has been no evaluation of how these models perform with in vitro genotoxicity data. Thus, the Genetic Toxicology Technical Committee, under the Health and Environmental Sciences Institute, conducted a case study on 31 reference chemicals to evaluate the performance of IVIVE application to genotoxicity data. The results demonstrate that for most chemicals considered here (20/31), the PODs derived from in vitro data and IVIVE are health protective relative to in vivo PODs from animal studies. PODs were also protective by assay target: mutations (8/13 chemicals), micronuclei (9/12), and aneugenicity markers (4/4). It is envisioned that this novel testing strategy could enhance prioritization, rapid screening, and risk assessment of genotoxic chemicals.

From vision toward best practices: Evaluating in vitro transcriptomic points of departure for application in risk assessment using a uniform workflow.

The growing number of chemicals in the current consumer and industrial markets presents a major challenge for regulatory programs faced with the need to assess the potential risks they pose to human and ecological health. The increasing demand for hazard and risk assessment of chemicals currently exceeds the capacity to produce the toxicity data necessary for regulatory decision making, and the applied data is commonly generated using traditional approaches with animal models that have limited context in terms of human relevance. This scenario provides the opportunity to implement novel, more efficient strategies for risk assessment purposes. This study aims to increase confidence in the implementation of new approach methods in a risk assessment context by using a parallel analysis to identify data gaps in current experimental designs, reveal the limitations of common approaches deriving transcriptomic points of departure, and demonstrate the strengths in using high-throughput transcriptomics (HTTr) to derive practical endpoints. A uniform workflow was applied across six curated gene expression datasets from concentration-response studies containing 117 diverse chemicals, three cell types, and a range of exposure durations, to determine tPODs based on gene expression profiles. After benchmark concentration modeling, a range of approaches was used to determine consistent and reliable tPODs. High-throughput toxicokinetics were employed to translate in vitro tPODs (µM) to human-relevant administered equivalent doses (AEDs, mg/kg-bw/day). The tPODs from most chemicals had AEDs that were lower (i.e., more conservative) than apical PODs in the US EPA CompTox chemical dashboard, suggesting in vitro tPODs would be protective of potential effects on human health. An assessment of multiple data points for single chemicals revealed that longer exposure duration and varied cell culture systems (e.g., 3D vs. 2D) lead to a decreased tPOD value that indicated increased chemical potency. Seven chemicals were flagged as outliers when comparing the ratio of tPOD to traditional POD, thus indicating they require further assessment to better understand their hazard potential. Our findings build confidence in the use of tPODs but also reveal data gaps that must be addressed prior to their adoption to support risk assessment applications.

Novel machine learning models to predict endocrine disruption activity for high-throughput chemical screening.

An area of ongoing concern in toxicology and chemical risk assessment is endocrine disrupting chemicals (EDCs). However, thousands of legacy chemicals lack the toxicity testing required to assess their respective EDC potential, and this is where computational toxicology can play a crucial role. The US (United States) Environmental Protection Agency (EPA) has run two programs, the Collaborative Estrogen Receptor Activity Project (CERAPP) and the Collaborative Modeling Project for Receptor Activity (CoMPARA) which aim to predict estrogen and androgen activity, respectively. The US EPA solicited research groups from around the world to provide endocrine receptor activity Qualitative (or Quantitative) Structure Activity Relationship ([Q]SAR) models and then combined them to create consensus models for different toxicity endpoints. Random Forest (RF) models were developed to cover a broader range of substances with high predictive capabilities using large datasets from CERAPP and CoMPARA for estrogen and androgen activity, respectively. By utilizing simple descriptors from open-source software and large training datasets, RF models were created to expand the domain of applicability for predicting endocrine disrupting activity and help in the screening and prioritization of extensive chemical inventories. In addition, RFs were trained to conservatively predict the activity, meaning models are more likely to make false-positive predictions to minimize the number of False Negatives. This work presents twelve binary and multi-class RF models to predict binding, agonism, and antagonism for estrogen and androgen receptors. The RF models were found to have high predictive capabilities compared to other in silico modes, with some models reaching balanced accuracies of 93% while having coverage of 89%. These models are intended to be incorporated into evolving priority-setting workflows and integrated strategies to support the screening and selection of chemicals for further testing and assessment by identifying potential endocrine-disrupting substances.

In Silico Models for Developmental Toxicity.

Modeling developmental toxicity has been a challenge for (Q)SAR model developers due to the complexity of the endpoint. Recently, some new in silico methods have been developed introducing the possibility to evaluate the integration of existing methods by taking advantage of various modeling perspectives. It is important that the model user is aware of the underlying basis of the different models in general, as well as the considerations and assumptions relative to the specific predictions that are obtained from these different models for the same chemical. The evaluation on the predictions needs to be done on a case-by-case basis, checking the analogues (possibly using structural, physicochemical, and toxicological information); for this purpose, the assessment of the applicability domain of the models provides further confidence in the model prediction. In this chapter, we present some examples illustrating an approach to combine human-based rules and statistical methods to support the prediction of developmental toxicity; we also discuss assumptions and uncertainties of the methodology.

The Effects of Organophosphate Esters Used as Flame Retardants and Plasticizers on Granulosa, Leydig, and Spermatogonial Cells Analyzed Using High-Content Imaging.

The replacement of regulated brominated flame retardants and plasticizers with organophosphate esters (OPEs) has led to their pervasive presence in the environment and in biological matrices. Further, there is evidence that exposure to some of these chemicals is associated with reproductive toxicity. Using a high-content imaging approach, we assessed the effects of exposure to 9 OPEs on cells related to reproductive function: KGN human granulosa cells, MA-10 mouse Leydig cells, and C18-4 mouse spermatogonial cells. The effects of OPEs were compared with those of 2,2',4,4'-tetrabromodiphenyl ether (BDE-47), a legacy brominated flame retardant. Alterations in several important cell features, including cell survival, mitochondrial dynamics, oxidative stress, lysosomes, and lipid droplets, were analyzed. Most of the OPEs tested displayed higher cytotoxicity than BDE-47 in all 3 cell lines. Effects on phenotypic parameters were specific for each cell type. Several OPEs increased total mitochondria, decreased lysosomes, increased the total area of lipid droplets, and induced oxidative stress in KGN cells; these endpoints were differentially affected in MA-10 and C18-4 cells. Alterations in cell phenotypes were highly correlated in the 2 steroidogenic cell lines for a few triaryl OPEs. Potency ranking using 2 complementary approaches, Toxicological Prioritization Index analyses and the lowest benchmark concentration/administered equivalent dose method, revealed that while most of the OPEs tested were more potent than BDE-47, others showed little to no effect. We propose that these approaches serve as lines of evidence in a screening strategy to identify the potential for reproductive and endocrine effects of emerging chemicals and assist in regulatory decision-making.

Implementing in vitro bioactivity data to modernize priority setting of chemical inventories.

Internationally, there are thousands of existing and newly introduced chemicals in commerce, highlighting the ongoing importance of innovative approaches to identify emerging chemicals of concern. For many chemicals, there is a paucity of hazard and exposure data. Thus, there is a crucial need for efficient and robust approaches to address data gaps and support risk-based prioritization. Several studies have demonstrated the utility of in vitro bioactivity data from the ToxCast program in deriving points of departure (PODs). ToxCast contains data for nearly 1,400 endpoints per chemical, and the bioactivity concentrations, indicative of potential adverse outcomes, can be converted to human-equivalent PODs using high-throughput toxicokinetics (HTTK) modeling. However, data gaps need to be addressed for broader application: the limited chemical space of HTTK and quantitative high-throughput screening data. Here we explore the applicability of in silico models to address these data needs. Specifically, we used ADMET predictor for HTTK predictions and a generalized read-across approach to predict ToxCast bioactivity potency. We applied these models to profile 5,801 chemicals on Canada's Domestic Substances List (DSL). To evaluate the approach's performance, bioactivity PODs were compared with in vivo results from the EPA Toxicity Values database for 1,042 DSL chemicals. Comparisons demonstrated that the bioac­tivity PODs, based on ToxCast data or read-across, were conservative for 95% of the chemicals. Comparing bioactivity PODs to human exposure estimates supports the identification of chemicals of potential interest for further work. The bioac­tivity workflow shows promise as a powerful screening tool to support effective triaging of chemical inventories.