Severity of effect considerations regarding the use of mutation as a toxicological endpoint for risk assessment: A report from the 8th International Workshop on Genotoxicity Testing (IWGT).

Exposure levels without appreciable human health risk may be determined by dividing a point of departure on a dose-response curve (e.g., benchmark dose) by a composite adjustment factor (AF). An "effect severity" AF (ESAF) is employed in some regulatory contexts. An ESAF of 10 may be incorporated in the derivation of a health-based guidance value (HBGV) when a "severe" toxicological endpoint, such as teratogenicity, irreversible reproductive effects, neurotoxicity, or cancer was observed in the reference study. Although mutation data have been used historically for hazard identification, this endpoint is suitable for quantitative dose-response modeling and risk assessment. As part of the 8th International Workshops on Genotoxicity Testing, a sub-group of the Quantitative Analysis Work Group (WG) explored how the concept of effect severity could be applied to mutation. To approach this question, the WG reviewed the prevailing regulatory guidance on how an ESAF is incorporated into risk assessments, evaluated current knowledge of associations between germline or somatic mutation and severe disease risk, and mined available data on the fraction of human germline mutations expected to cause severe disease. Based on this review and given that mutations are irreversible and some cause severe human disease, in regulatory settings where an ESAF is used, a majority of the WG recommends applying an ESAF value between 2 and 10 when deriving a HBGV from mutation data. This recommendation may need to be revisited in the future if direct measurement of disease-causing mutations by error-corrected next generation sequencing clarifies selection of ESAF values.

Visualization strategies to aid interpretation of high-dimensional genotoxicity data.

This article describes a range of high-dimensional data visualization strategies that we have explored for their ability to complement machine learning algorithm predictions derived from MultiFlow® assay results. For this exercise, we focused on seven biomarker responses resulting from the exposure of TK6 cells to each of 126 diverse chemicals over a range of concentrations. Obviously, challenges associated with visualizing seven biomarker responses were further complicated whenever there was a desire to represent the entire 126 chemical data set as opposed to results from a single chemical. Scatter plots, spider plots, parallel coordinate plots, hierarchical clustering, principal component analysis, toxicological prioritization index, multidimensional scaling, t-distributed stochastic neighbor embedding, and uniform manifold approximation and projection are each considered in turn. Our report provides a comparative analysis of these techniques. In an era where multiplexed assays and machine learning algorithms are becoming the norm, stakeholders should find some of these visualization strategies useful for efficiently and effectively interpreting their high-dimensional data.

An integrated in vitro carcinogenicity test that distinguishes between genotoxic carcinogens, non-genotoxic carcinogens, and non-carcinogens.

Chemical safety testing plays a crucial role in product and pharmacological development, as well as chemoprevention; however, in vitro genotoxicity safety tests do not always accurately predict the chemicals that will be in vivo carcinogens. If chemicals test positive in vitro for genotoxicity but negative in vivo, this can contribute to unnecessary testing in animals used to confirm erroneous in vitro positive results. Current in vitro tests typically evaluate only genotoxicity endpoints, which limits their potential to detect non-genotoxic carcinogens. The frequency of misleading in vitro positive results can be high, leading to a requirement for more informative in vitro tests. It is now recognized that multiple-endpoint genotoxicity testing may aid more accurate detection of carcinogens and non-carcinogens. The objective of this review was to evaluate the utility of our novel, multiple-endpoint in vitro test, which uses multiple cancer-relevant endpoints to predict carcinogenic potential. The tool assessed micronucleus frequency, p53 expression, p21 expression, mitochondrial respiration, cell cycle abnormalities and, uniquely, cell morphology changes in human lymphoblastoid cell lines, TK6 and MCL-5. The endpoints were used to observe cellular responses to 18 chemicals within the following categories: genotoxic carcinogens, non-genotoxic carcinogens, toxic non-carcinogens, and misleading in vitro positive and negative agents. The number of endpoints significantly altered for each chemical was considered, alongside the holistic Integrated Signature of Carcinogenicity score, derived from the sum of fold changes for all endpoints. Following the calculation of an overall score from these measures, carcinogens exhibited greater potency than non-carcinogens. Genotoxic carcinogens were generally more potent than non-genotoxic carcinogens. This novel approach therefore demonstrated potential for correctly predicting whether chemicals with unknown mechanism may be considered carcinogens. Overall, while further validation is recommended, the test demonstrates potential for the identification of carcinogenic compounds. Adoption of the approach could enable reduced animal use in carcinogenicity testing.

Nitrosamine acceptable intakes should consider variation in molecular weight: The implication of stoichiometric DNA damage.

N-nitrosamines (NAs) are a class of compounds of which many, especially of the small dialkyl type, are indirect acting DNA alkylating mutagens. Their presence in pharmaceuticals is subject to very strict acceptable daily intake (AI) limits, which are traditionally expressed on a mass basis. Here we demonstrate that AIs that are not experimentally derived for a specific compound, but via statistical extrapolation or read across to a suitable analog, should be expressed on a molar scale or corrected for the target substance's molecular weight. This would account for the mechanistic aspect that each nitroso group can, at maximum, account for a single DNA mutation and the number of molecules per mass unit is proportional to the molecular weight (MW). In this regard we have re-calculated the EMA 18 ng/day regulatory default AI for unknown nitrosamines on a molar scale and propose a revised default AI of 163 pmol/day. In addition, we provide MW-corrected AIs for those nitrosamine drug substance related impurities (NDSRIs) for which EMA has pre-assigned AIs by read-across. Regulatory acceptance of this fundamental scientific tenet would allow one to derive nitrosamine limits for NDSRIs that both meet the health-protection goals and are technically feasible.

Inter-laboratory automation of the in vitro micronucleus assay using imaging flow cytometry and deep learning.

The in vitro micronucleus assay is a globally significant method for DNA damage quantification used for regulatory compound safety testing in addition to inter-individual monitoring of environmental, lifestyle and occupational factors. However, it relies on time-consuming and user-subjective manual scoring. Here we show that imaging flow cytometry and deep learning image classification represents a capable platform for automated, inter-laboratory operation. Images were captured for the cytokinesis-block micronucleus (CBMN) assay across three laboratories using methyl methanesulphonate (1.25-5.0 µg/mL) and/or carbendazim (0.8-1.6 µg/mL) exposures to TK6 cells. Human-scored image sets were assembled and used to train and test the classification abilities of the "DeepFlow" neural network in both intra- and inter-laboratory contexts. Harnessing image diversity across laboratories yielded a network able to score unseen data from an entirely new laboratory without any user configuration. Image classification accuracies of 98%, 95%, 82% and 85% were achieved for 'mononucleates', 'binucleates', 'mononucleates with MN' and 'binucleates with MN', respectively. Successful classifications of 'trinucleates' (90%) and 'tetranucleates' (88%) in addition to 'other or unscorable' phenotypes (96%) were also achieved. Attempts to classify extremely rare, tri- and tetranucleated cells with micronuclei into their own categories were less successful (≤ 57%). Benchmark dose analyses of human or automatically scored micronucleus frequency data yielded quantitation of the same equipotent concentration regardless of scoring method. We conclude that this automated approach offers significant potential to broaden the practical utility of the CBMN method across industry, research and clinical domains. We share our strategy using openly-accessible frameworks.

Permitted daily exposure limits for noteworthy N-nitrosamines.

A genotoxic carcinogen, N-nitrosodimethylamine (NDMA), was detected as a synthesis impurity in some valsartan drugs in 2018, and other N-nitrosamines, such as N-nitrosodiethylamine (NDEA), were later detected in other sartan products. N-nitrosamines are pro-mutagens that can react with DNA following metabolism to produce DNA adducts, such as O6 -alkyl-guanine. The adducts can result in DNA replication miscoding errors leading to GC>AT mutations and increased risk of genomic instability and carcinogenesis. Both NDMA and NDEA are known rodent carcinogens in male and female rats. The DNA repair enzyme, methylguanine DNA-methyltransferase can restore DNA integrity via the removal of alkyl groups from guanine in an error-free fashion and this can result in nonlinear dose responses and a point of departure or "practical threshold" for mutation at low doses of exposure. Following International recommendations (ICHM7; ICHQ3C and ICHQ3D), we calculated permissible daily exposures (PDE) for NDMA and NDEA using published rodent cancer bioassay and in vivo mutagenicity data to determine benchmark dose values and define points of departure and adjusted with appropriate uncertainty factors (UFs). PDEs for NDMA were 6.2 and 0.6 µg/person/day for cancer and mutation, respectively, and for NDEA, 2.2 and 0.04 µg/person/day. Both PDEs are higher than the acceptable daily intake values (96 ng for NDMA and 26.5 ng for NDEA) calculated by regulatory authorities using simple linear extrapolation from carcinogenicity data. These PDE calculations using a bench-mark approach provide a more robust assessment of exposure limits compared with simple linear extrapolations and can better inform risk to patients exposed to the contaminated sartans.

Empirical comparison of genotoxic potency estimations: the in vitro DNA-damage ToxTracker endpoints versus the in vivo micronucleus assay.

Genetic toxicology is an essential component of compound safety assessment. In the face of a barrage of new compounds, higher throughput, less ethically divisive in vitro approaches capable of effective, human-relevant hazard identification and prioritisation are increasingly important. One such approach is the ToxTracker assay, which utilises murine stem cell lines equipped with green fluorescent protein (GFP)-reporter gene constructs that each inform on distinct aspects of cellular perturbation. Encouragingly, ToxTracker has shown improved sensitivity and specificity for the detection of known in vivo genotoxicants when compared to existing 'standard battery' in vitro tests. At the current time however, quantitative genotoxic potency correlations between ToxTracker and well-recognised in vivo tests are not yet available. Here we use dose-response data from the three DNA-damage-focused ToxTracker endpoints and from the in vivo micronucleus assay to carry out quantitative, genotoxic potency estimations for a range of aromatic amine and alkylating agents using the benchmark dose (BMD) approach. This strategy, using both the exponential and the Hill BMD model families, was found to produce robust, visually intuitive and similarly ordered genotoxic potency rankings for 17 compounds across the BSCL2-GFP, RTKN-GFP and BTG2-GFP ToxTracker endpoints. Eleven compounds were similarly assessed using data from the in vivo micronucleus assay. Cross-systems genotoxic potency correlations for the eight matched compounds demonstrated in vitro-in vivo correlation, albeit with marked scatter across compounds. No evidence for distinct differences in the sensitivity of the three ToxTracker endpoints was found. The presented analyses show that quantitative potency determinations from in vitro data enable more than just qualitative screening and hazard identification in genetic toxicology.

Tolerability of risk: A commentary on the nitrosamine contamination issue.

For decades, regulators have grappled with different approaches to address the issue of control of impurities. Safety-based limits, such as permissible daily exposure (PDE), acceptable intake (AI), threshold of toxicological concern (TTC) and less than lifetime limits (LTL) have all been used. For many years these safety-based limits have been recognized as virtually safe doses (VSDs). Recently, however, many regulatory agencies are seeking to impose limits for N-nitrosamine impurities, which are significantly below the VSD. This commentary will discuss the evolution of safety-based limits for impurities, provide an overview of the valsartan N-nitrosamine contamination issue and review the toxicology of N-nitrosamines. The outcome of a lessons-learned exercise on sartan medications undertaken by the European Medicines Agency (EMA) will also be discussed. The review will also highlight the many analytical challenges inherent with controlling impurities to ppb-based limits. The use of highly sensitive, low ppb limits, methods may lead to future issues of batch rejection, based on false positives. Regulators initially viewed the N-nitrosamine risk as being insufficient to prompt immediate product discontinuation and patients were specifically advised to continue using their affected medication. Patients were also informed that exposure to N-nitrosamines is extremely common via food and drinking water.

The use of benchmark dose uncertainty measurements for robust comparative potency analyses.

The Benchmark Dose (BMD) method is the favored approach for quantitative dose-response analysis where uncertainty measurements are delineated between the upper (BMDU) and lower (BMDL) confidence bounds, or confidence intervals (CIs). Little has been published on the accurate interpretation of uncertainty measurements for potency comparative analyses between different test conditions. We highlight this by revisiting a previously published comparative in vitro genotoxicity dataset for human lymphoblastoid TK6 cells that were exposed to each of 10 clastogens in the presence and absence (+/-) of low concentration (0.25%) S9, and scored for p53, γH2AX and Relative Nuclei Count (RNC) responses at two timepoints (Tian et al., 2020). The researchers utilized BMD point estimates in potency comparative analysis between S9 treatment conditions. Here we highlight a shortcoming that the use of BMD point estimates can mischaracterize potency differences between systems. We reanalyzed the dose responses by BMD modeling using PROAST v69.1. We used the resulting BMDL and BMDU metrics to calculate "S9 potency ratio confidence intervals" that compare the relative potency of compounds +/- S9 as more statistically robust metrics for comparative potency measurements compared to BMD point estimate ratios. We performed unsupervised hierarchical clustering that identified four S9-dependent groupings: high and low-level potentiation, no effect, and diminution. This work demonstrates the importance of using BMD uncertainty measurements in potency comparative analyses between test conditions. Irrespective of the source of the data, we propose a stepwise approach when performing BMD modeling in comparative potency analyses between test conditions.

Multiple-endpoint in vitro carcinogenicity test in human cell line TK6 distinguishes carcinogens from non-carcinogens and highlights mechanisms of action.

Current in vitro genotoxicity tests can produce misleading positive results, indicating an inability to effectively predict a compound's subsequent carcinogenic potential in vivo. Such oversensitivity can incur unnecessary in vivo tests to further investigate positive in vitro results, supporting the need to improve in vitro tests to better inform risk assessment. It is increasingly acknowledged that more informative in vitro tests using multiple endpoints may support the correct identification of carcinogenic potential. The present study, therefore, employed a holistic, multiple-endpoint approach using low doses of selected carcinogens and non-carcinogens (0.001-770 µM) to assess whether these chemicals caused perturbations in molecular and cellular endpoints relating to the Hallmarks of Cancer. Endpoints included micronucleus induction, alterations in gene expression, cell cycle dynamics, cell morphology and bioenergetics in the human lymphoblastoid cell line TK6. Carcinogens ochratoxin A and oestradiol produced greater Integrated Signature of Carcinogenicity scores for the combined endpoints than the "misleading" in vitro positive compounds, quercetin, 2,4-dichlorophenol and quinacrine dihydrochloride and toxic non-carcinogens, caffeine, cycloheximide and phenformin HCl. This study provides compelling evidence that carcinogens can successfully be distinguished from non-carcinogens using a holistic in vitro test system. Avoidance of misleading in vitro outcomes could lead to the reduction and replacement of animals in carcinogenicity testing.

A weight of evidence assessment of the genotoxicity of 2,6-xylidine based on existing and new data, with relevance to safety of lidocaine exposure.

Lidocaine has not been associated with cancer in humans despite 8 decades of therapeutic use. Its metabolite, 2,6-xylidine, is a rat carcinogen, believed to induce genotoxicity via N-hydroxylation and DNA adduct formation, a non-threshold mechanism of action. To better understand this dichotomy, we review literature pertaining to metabolic activation and genotoxicity of 2,6-xylidine, identifying that it appears resistant to N-hydroxylation and instead metabolises almost exclusively to DMAP (an aminophenol). At high exposures (sufficient to saturate phase 2 metabolism), this may undergo metabolic threshold-dependent activation to a quinone-imine with potential to redox cycle producing ROS, inducing cytotoxicity and genotoxicity. A new rat study found no evidence of genotoxicity in vivo based on micronuclei in bone marrow, comets in nasal tissue or female liver, despite high level exposure to 2,6-xylidine (including metabolites). In male liver, weak dose-related comet increases, within the historical control range, were associated with metabolic overload and acute systemic toxicity. Benchmark dose analysis confirmed a non-linear dose response. The weight of evidence indicates 2,6-xylidine is a non-direct acting (metabolic threshold-dependent) genotoxin, and is not genotoxic in vivo in rats in the absence of acute systemic toxic effects, which occur at levels 35 × beyond lidocaine-related exposure in humans.

Detection of urethane-induced genotoxicity in vitro using metabolically competent human 2D and 3D spheroid culture models.

In vitro genotoxicity studies are a quick and high throughput approach to assess the genotoxic potential of chemicals; however, the reliability of these tests and their relevance to in vivo effects depends on the choice of representative cell line and optimisation of assay conditions. For chemicals like urethane that require specific metabolic activation to cause genotoxicity, it is important that in vitro tests are conducted using cell lines exhibiting the activity and induction of CYP450 enzymes, including CYP2E1 enzyme that is important in the metabolism of urethane, at a concentration representing actual or perceived chemical exposure. We compared 2D MCL-5 cells and HepG2 cells with 3D HepG2 hanging drop spheroids to determine the genotoxicity of urethane using the micronucleus assay. Our 2D studies with MCL-5 did not show any statistically significant genotoxicity [99% relative population doubling (RPD)] compared to controls for concentrations and time point tested in vitro. HepG2 cells grown as 2D indicated that exposure to urethane of up to 30 mM for 23 h did not cause any genotoxic effect (102% RPD) but, at higher concentrations, genotoxicity was produced with only 89-85% RPD. Furthermore, an exposure of 20-50 mM for 23 h using 3D hanging drop spheroid assays revealed a higher MN frequency, thus exhibiting in vitro genotoxicity of urethane in metabolically active cell models. In comparison with previous studies, this study indicated that urethane genotoxicity is dose, sensitivity of cell model (2D vs. 3D) and exposure dependent.

Genotoxicity as a toxicologically relevant endpoint to inform risk assessment: A case study with ethylene oxide.

The purpose of the present investigation is to analyze the in vivo genotoxicity dose-response data of ethylene oxide (EO) and the applicability of the derived point-of-departure (PoD) values when estimating permitted daily exposure (PDE) values. A total of 40 data sets were identified from the literature, and benchmark dose analyses were conducted using PROAST software to identify a PoD value. Studies employing the inhalation route of exposure and assessing gene or chromosomal mutations and chromosomal damage in various tissues were considered the most relevant for assessing risk from EO, since these effects are likely to contribute to adverse health consequences in exposed individuals. The PoD estimates were screened for precision and the values were divided by data-derived adjustment factors. For gene mutations, the lowest PDE was 285 parts per trillion (ppt) based on the induction of lacI mutations in the testes of mice following 48 weeks of exposure to EO. The corresponding lowest PDE value for chromosomal mutations was 1,175 ppt for heritable translocations in mice following 8.5 weeks of EO exposure. The lowest PDE for chromosomal aberrations was 238 ppt in the mouse peripheral blood lymphocytes following 48 weeks of inhalation exposure. The diverse dose-response data for EO-induced genotoxicity enabled the derivation of PoDs for various endpoints, tissues, and species and identified 238 ppt as the lowest PDE in this retrospective analysis.

Benchmark Dose Analysis of DNA Damage Biomarker Responses Provides Compound Potency and Adverse Outcome Pathway Information for the Topoisomerase II Inhibitor Class of Compounds.

Genetic toxicology data have traditionally been utilized for hazard identification to provide a binary call for a compound's risk. Recent advances in the scientific field, especially with the development of high-throughput methods to quantify DNA damage, have influenced a change of approach in genotoxicity assessment. The in vitro MultiFlow® DNA Damage Assay is one such method which multiplexes γH2AX, p53, phospho-histone H3 biomarkers into a single-flow cytometric analysis (Bryce et al., [2016]: Environ Mol Mutagen 57:546-558). This assay was used to study human TK6 cells exposed to each of eight topoisomerase II poisons for 4 and 24 hr. Using PROAST v65.5, the Benchmark Dose approach was applied to the resulting flow cytometric datasets. With "compound" serving as covariate, all eight compounds were combined into a single analysis, per time point and endpoint. The resulting 90% confidence intervals, plotted in Log scale, were considered as the potency rank for the eight compounds. The in vitro MultiFlow data showed a maximum confidence interval span of 1Log, which indicates data of good quality. Patterns observed in the compound potency rank were scrutinized by using the expert rule-based software program Derek Nexus, developed by Lhasa Limited. Compound sub-classification and structural alerts were considered contributory to the potencies observed for the topoisomerase II poisons studied herein. The Topo II poison Adverse Outcome Pathway was evaluated with MultiFlow endpoints serving as Key Events. The step-wise approach described herein can be considered as a foundation for risk assessment of compounds within a specific mode of action of interest. Environ. Mol. Mutagen. 2020. © 2020 Wiley Periodicals, Inc.

Quantitative Interpretation of Genetic Toxicity Dose-Response Data for Risk Assessment and Regulatory Decision-Making: Current Status and Emerging Priorities.

The screen-and-bin approach for interpretation of genotoxicity data is predicated on three false assumptions: that genotoxicants are rare, that genotoxicity dose-response functions do not contain a low-dose region mechanistically characterized by zero-order kinetics, and that genotoxicity is not a bona fide toxicological endpoint. Consequently, there is a need to develop and implement quantitative methods to interpret genotoxicity dose-response data for risk assessment and regulatory decision-making. Standardized methods to analyze dose-response data, and determine point-of-departure (PoD) metrics, have been established; the most robust PoD is the benchmark dose (BMD). However, there are no standards for regulatory interpretation of mutagenicity BMDs. Although 5-10% is often used as a critical effect size (CES) for BMD determination, values for genotoxicity endpoints have not been established. The use of BMDs to determine health-based guidance values (HBGVs) requires assessment factors (AFs) to account for interspecies differences and variability in human sensitivity. Default AFs used for other endpoints may not be appropriate for interpretation of in vivo mutagenicity BMDs. Analyses of published dose-response data showing the effects of compensatory pathway deficiency indicate that AFs for sensitivity differences should be in the range of 2-20. Additional analyses indicate that the AF to compensate for short treatment durations should be in the range of 5-15. Future work should use available data to empirically determine endpoint-specific CES values; similarly, to determine AF values for BMD adjustment. Future work should also evaluate the ability to use in vitro dose-response data for risk assessment, and the utility of probabilistic methods for determination of mutagenicity HBGVs. Environ. Mol. Mutagen. 61:66-83, 2020. © 2019 Her Majesty the Queen in Right of Canada.

Mutation as a Toxicological Endpoint for Regulatory Decision-Making.

Mutations induced in somatic cells and germ cells are responsible for a variety of human diseases, and mutation per se has been considered an adverse health concern since the early part of the 20th Century. Although in vitro and in vivo somatic cell mutation data are most commonly used by regulatory agencies for hazard identification, that is, determining whether or not a substance is a potential mutagen and carcinogen, quantitative mutagenicity dose-response data are being used increasingly for risk assessments. Efforts are currently underway to both improve the measurement of mutations and to refine the computational methods used for evaluating mutation data. We recommend continuing the development of these approaches with the objective of establishing consensus regarding the value of including the quantitative analysis of mutation per se as a required endpoint for comprehensive assessments of toxicological risk. Environ. Mol. Mutagen. 61:34-41, 2020. © 2019 Wiley Periodicals, Inc.

Utility of a next generation framework for assessment of genomic damage: A case study using the industrial chemical benzene.

We recently published a next generation framework for assessing the risk of genomic damage via exposure to chemical substances. The framework entails a systematic approach with the aim to quantify risk levels for substances that induce genomic damage contributing to human adverse health outcomes. Here, we evaluated the utility of the framework for assessing the risk for industrial chemicals, using the case of benzene. Benzene is a well-studied substance that is generally considered a genotoxic carcinogen and is known to cause leukemia. The case study limits its focus on occupational and general population health as it relates to benzene exposure. Using the framework as guidance, available data on benzene considered relevant for assessment of genetic damage were collected. Based on these data, we were able to conduct quantitative analyses for relevant data sets to estimate acceptable exposure levels and to characterize the risk of genetic damage. Key observations include the need for robust exposure assessments, the importance of information on toxicokinetic properties, and the benefits of cheminformatics. The framework points to the need for further improvement on understanding of the mechanism(s) of action involved, which would also provide support for the use of targeted tests rather than a prescribed set of assays. Overall, this case study demonstrates the utility of the next generation framework to quantitatively model human risk on the basis of genetic damage, thereby enabling a new, innovative risk assessment concept. Environ. Mol. Mutagen. 61:94-113, 2020. © 2019 The Authors. Environmental and Molecular Mutagenesis published by Wiley Periodicals, Inc. on behalf of Environmental Mutagen Society.

Correction to: Re: Gi et al. 2018, In vivo positive mutagenicity of 1,4-dioxane and quantitative analysis of its mutagenicity and carcinogenicity in rats, Archives of Toxicology 92:3207-3221.

The publisher would like to apologize for the failed cross-linking to the following Letter to the Editor by Paul A.

A comparison of transgenic rodent mutation and in vivo comet assay responses for 91 chemicals.

A database of 91 chemicals with published data from both transgenic rodent mutation (TGR) and rodent comet assays has been compiled. The objective was to compare the sensitivity of the two assays for detecting genotoxicity. Critical aspects of study design and results were tabulated for each dataset. There were fewer datasets from rats than mice, particularly for the TGR assay, and therefore, results from both species were combined for further analysis. TGR and comet responses were compared in liver and bone marrow (the most commonly studied tissues), and in stomach and colon evaluated either separately or in combination with other GI tract segments. Overall positive, negative, or equivocal test results were assessed for each chemical across the tissues examined in the TGR and comet assays using two approaches: 1) overall calls based on weight of evidence (WoE) and expert judgement, and 2) curation of the data based on a priori acceptability criteria prior to deriving final tissue specific calls. Since the database contains a high prevalence of positive results, overall agreement between the assays was determined using statistics adjusted for prevalence (using AC1 and PABAK). These coefficients showed fair or moderate to good agreement for liver and the GI tract (predominantly stomach and colon data) using WoE, reduced agreement for stomach and colon evaluated separately using data curation, and poor or no agreement for bone marrow using both the WoE and data curation approaches. Confidence in these results is higher for liver than for the other tissues, for which there were less data. Our analysis finds that comet and TGR generally identify the same compounds (mainly potent mutagens) as genotoxic in liver, stomach and colon, but not in bone marrow. However, the current database content precluded drawing assay concordance conclusions for weak mutagens and non-DNA reactive chemicals.

Predictions of genotoxic potential, mode of action, molecular targets, and potency via a tiered multiflow® assay data analysis strategy.

The in vitro MultiFlow® DNA Damage Assay multiplexes γH2AX, p53, phospho-histone H3, and polyploidization biomarkers into a single flow cytometric analysis. The current report describes a tiered sequential data analysis strategy based on data generated from exposure of human TK6 cells to a previously described 85 chemical training set and a new pharmaceutical-centric test set (n = 40). In each case, exposure was continuous over a range of closely spaced concentrations, and cell aliquots were removed for analysis following 4 and 24 hr of treatment. The first data analysis step focused on chemicals' genotoxic potential, and for this purpose, we evaluated the performance of a machine learning (ML) ensemble, a rubric that considered fold increases in biomarkers against global evaluation factors (GEFs), and a hybrid strategy that considered ML and GEFs. This first tier further used ML output and/or GEFs to classify genotoxic activity as clastogenic and/or aneugenic. Test set results demonstrated the generalizability of the first tier, with particularly good performance from the ML ensemble: 35/40 (88%) concordance with a priori genotoxicity expectations and 21/24 (88%) agreement with expected mode of action (MoA). A second tier applied unsupervised hierarchical clustering to the biomarker response data, and these analyses were found to group certain chemicals, especially aneugens, according to their molecular targets. Finally, a third tier utilized benchmark dose analyses and MultiFlow biomarker responses to rank genotoxic potency. The relevance of these rankings is supported by the strong agreement found between benchmark dose values derived from MultiFlow biomarkers compared to those generated from parallel in vitro micronucleus analyses. Collectively, the results suggest that a tiered MultiFlow data analysis pipeline is capable of rapidly and effectively identifying genotoxic hazards while providing additional information that is useful for modern risk assessments-MoA, molecular targets, and potency. Environ. Mol. Mutagen. 60:513-533, 2019. © 2019 Wiley Periodicals, Inc.