EUROTOX 2018 ControlCenter

Online Program Overview Session: SOC 2

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

Short oral communications 2

Session chair: François Huaux Belgium
Shortcut: SOC 2
Date: Tuesday, 4 September, 2018, 10:00
Room: Hall 300
Session type: Short Oral Communication


Click on an contribution to preview the abstract content.


Novel Computational Tools based on Bioinformatic and Chemoinformatic Data to Complement Zebrafish Embryo Teratogenicity Test (#38)

C. Quevedo1, M. Ipiñazar1, F. J. Planes2, X. Cendoya2

1 Biobide, San Sebastian, Spain
2 Tecnun, University of Navarra, San Sebastian, Spain

Approximately 3% of newborns present congenital anomalies and around 5-10% of those are caused by exposure to teratogenic agents [1]. For this reason, regulatory organisms and the industry demand for effective methods to test the developmental toxicity of drugs, industry chemicals or waste products [2]. The use of the zebrafish embryotoxicity test is an attractive strategy to minimize in-vivo assays and animal models. Overall, this assay has a good predictability; however, the outcome is based on morphologic evaluation, which is subjective and subtle effects might be neglected [3]. With the increasing amount of molecular databases, the development of in-silico tools that complement experimental assays is promising. In this work, we present an in-silico platform that makes use of bioinformatics and chemoinformatics data, as well as machine learning methods, in order to predict the teratogenic potential of a particular compound. First, we show a combined systems biology and metabolomics study in order to identify metabolic biomarkers that improve the sensitivity of the zebrafish embryotoxicity test. Second, a learning algorithm using structural information is evaluated and compared using publicly available data, analyzing their complementarity with the zebrafish embryotoxicity test and metabolic biomarkers with newly generated proprietary data.


1. Seiler A, Visan A, Buesen R, et al. The use of embryonic stem cells for developmental toxicity testing. Reproductive Toxicology 2009; 28:141–142
2. Yang L, Kemadjou JR, Zinsmeister C, et al. Transcriptional profiling reveals barcode-like toxicogenomic responses in the zebrafish embryo. Genome Biol 2007; 8:R227
3. Hermsen SA, Pronk TE, Brandhof E-J van den, et al. Transcriptomic analysis in the developing zebrafish embryo after compound exposure: individual gene expression and pathway regulation. Toxicology and applied pharmacology 2013; 272:161–171

Keywords: developmental toxicity, zebrafish, in silico, embryotoxicity, alternative animal models, metabolic biomarkers, bioinformatics, chemoinformatics

Prediction of Adverse Male Reproductive Health Effects by integrating In Vitro Data and Physiologically-Based Kinetic Modelling (#72)

C. Taxvig1, M. Scholze2, A. Kortenkamp2, J. Boberg1, T. Svingen1, S. Ermler2, S. S. Hermann1, A. K. Lykkeberg1, M. Pedersen1, U. Hass1, S. Bonomo1, A. M. Vinggaard1

1 Technical University of Denmark, National Food Institute, Kgs.Lyngby, Denmark
2 Brunel University, London, United Kingdom

Currently, ~352 pesticides are approved within the EU, but knowledge concerning sensitive endocrine effects on male reproductive health is scarce. Thus, there is an urgent need to improve non‑animal test strategies that can help in predicting pesticides for adverse effects on male reproductive health.

The development of the male reproductive system strongly depends on androgens produced by the fetal testes, and hence the fetus is a target for compounds capable of interfering with the synthesis of these hormones or by antagonizing the androgen receptor.

Our strategy combines androgen‑related activity of pesticides on human cells with physiologically-based kinetic modeling. Here, in vitro data pinpoints to compounds with a potential in vivo activity by identifying their critical internal exposure, while the kinetic model simulates the maternal doses necessary to reach critical levels in the fetus. We have developed a proof of principle showing that adverse effects on anogenital distance in male offspring, which is a unique and non‑invasive marker for male reproductive health effects in animals and humans, can be predicted for selected pesticides.

We investigated this strategy on 9 pesticides and selected 3 compounds – fludioxonil, cyprodinil and dimethomorph – for in vivo ‘validation’ of the alternative approach in rats. Predicted fetal levels were within a factor of 2 from measured concentrations, and all three compounds showed a shorthened AGD in vivo.

In conclusion, our approach can be used to avoid unnecessary animal testing and focus on compounds that most likely will produce in vivo activity.

Keywords: Male reproduction, PBK modelling, in vitro assays, anogenital distance

Troglitazone and Rosiglitazone profiles in Tox21/Toxcast: verifying results with toxicology data from published in vivo data and human pharmacovigilance data using an evidence-based approach (#97)

H. Dirven1, G. Vist2, A. Sayed3, S. Bandhakavi4, M. Branik3, N. Peace4, K. Tsaioun5

1 Norwegian Institute of Public Health, Department of Toxicology and Risk Assessment, Oslo, Norway
2 Norwegian Institute of Public Health, Division of Health Services, Oslo, Norway
3 Douglas Connect GmbH, Basel, Switzerland
4 Northeastern University, Boston, United States of America
5 Evidence-Based Toxicology Consortium, Baltimore, United States of America

In order to study the ability of in vitro data to predict the hazards of chemicals, we analysed the in vitro data from the USA EPA ToxCastTM program on Troglitazone and Rosiglitazone, and compared it with the profile obtained from a systematic literature review on the results of standard toxicity tests in experimental animals and adverse findings as reported in patients. In addition, findings with the 2 drugs as found in Pharmacovigilance databases are summarised.

Both drugs are antidiabetic drugs: Troglitazone was in clinical practice associated with drug-induced hepatitis, while Rosiglitazone was linked to increased risk of heart attack.

Activity in Tox21/Toxcast (EC50 values, normalised to the human Cmax values), showed that Troglitazone had activity in 50% more assays compared to Rosiglitazone.

A protocol-driven systematic literature review on liver-associated effects of the two drugs in dogs, mice, rats, non-human primates and humans was performed. Studies in experimental disease models or covering multiple treatments were excluded. In total 105 abstracts for Troglitazone and 175 abstracts for Rosiglitazone were found. After a second screening based on full-text papers, at least 88 papers on Rosiglitazone and 48 papers on Troglitazone are selected for limited data extraction. A profile of adverse liver effects of each drug as reported in these papers will be presented.

To incorporate real-world data on frequencies of adverse effects in patients, the WHO VigiBase pharmacovigilance database was analysed for liver-related adverse events. Adverse effects  in the liver  of each drug as extracted from this database  will be summarised.

Results of these 3 evidence streams will be presented and compared in order to judge if the results as obtained in Tox21/Toxcast are predictive for effects observed in experimental animals and humans. Challenges with analysing the data using evidence-based methods will be described.

Keywords: Tox21/Toxcast, Troglitazone, Rosiglitazone, Systematic review

Use of Chemical-specific Adjustment Factors in Health Risk Assessments: A Tale of Two Preservatives (#175)

J. E. Goodman1, D. M. Pizzurro1, J. C. Lemay1, T. A. Lewandowski2, K. Zu1

1 Gradient, Cambridge, Massachusetts, United States of America
2 Gradient, Seattle, Washington, United States of America

Regulatory bodies are increasingly using chemical-specific adjustment factors (CSAFs) instead of default uncertainty factors (UFs) to derive acceptable daily intakes (ADIs) and determine margins of safety (MoSs) for substances in consumer products, pharmaceuticals, and the environment. Here, we demonstrate how the use of CSAFs reduced uncertainty in chemical risk assessments in two case studies. First, we evaluated benzoic acid and its sodium and potassium salts, which are common preservatives in foods and beverages and have a group ADI of 0-5 mg/kg bw-day based on a default total UF of 100. Comparing chemical-specific pharmacokinetic data (e.g., areas under the curve) in rodents and humans, we derived a CSAF of 2 for the pharmacokinetic component of the interspecies UF (i.e., PKinterCSAF), which would increase the ADI to 0-10 mg/kg bw-day. For the intraspecies UF, we conducted an exploratory analysis to compare the maximum concentrations in blood in adults and children and found that the pharmacokinetic component of this UF could be reduced from the default value of 3.16, suggesting that the ADI could be increased further. Second, current regulations indicate that personal care or cosmetic products should not contain more than 1% phenoxyethanol, a common preservative. Physiologically based pharmacokinetic modeling and comparison of chemical-specific data in rodents and humans both support a PKinterCSAF of 1, resulting in a minimum MoS of 25 instead of the default 100, when extrapolating from animal data to humans. We assessed whether this regulatory limit provides an adequate MoS for infants and small children using several personal care products containing phenoxyethanol. We first conducted a hazard assessment and identified points of departure (PODs) from animal toxicity studies. We then estimated children's exposures to phenoxyethanol from using several moisturizing and bathing products. Comparing the exposure estimates to the PODs, we calculated MoSs for infants and small children of 210 and 1,660 for dermal and oral exposures, respectively, both of which are considerably greater than the minimum MoS of 25. In both case studies, the use of data-driven CSAFs reduced the uncertainty in chemical risk assessments and enhanced confidence in the safety of these components in real-world conditions.

Keywords: Chemical-specific adjustment factor, risk assessment, sodium benzoate, phenoxyethanol, preservative

Deriving a no expected sensitization induction level for fragrance ingredients without animal testing: Specific case studies assessed with different approaches (#198)

A. Natsch1, R. Emter1, T. Haupt1, G. Ellis2

1 Givaudan Schweiz AG, Ingredients Research, Duebendorf, Switzerland
2 Givaudan International SA, Regulatory Affairs and Product Safety, Vernier, Switzerland

Cosmetic regulations prohibit animal testing for the purpose of safety assessment and recent REACH guidance states that the local lymph node assay (LLNA) in mice shall only be conducted if in vitro data cannot give sufficient information for classification and labelling. However, Quantitative Risk Assessment (QRA) for fragrance ingredients requires a NESIL, a dose not expected to cause induction of skin sensitization in humans. In absence of human data, this is derived from the LLNA and it remains a key challenge for risk assessors to derive this value from non-animal data. Here we present a workflow using structural information, reactivity data and KeratinoSens results to predict a LLNA result as a point of departure. Specific additional tests (metabolic activation, complementary reactivity tests) are applied in selected cases depending on the chemical domain of a molecule. Most importantly, in vitro and in vivo data on close analogues are used to estimate uncertainty of the prediction in the specific chemical domain. This approach was applied to three molecules which were subsequently tested in the LLNA and 22 molecules with available and sometimes discordant human and LLNA data. Four additional case studies illustrate how this approach is being applied to recently developed molecules in the absence of animal data. We then further compared, on the same molecules, this approach with other published data integration strategies also using additional data sources (TIMES-SS in silico prediction, human cell line activation test). For most molecules, similar results are obtained with different strategies, which can be explained by partly redundant data generated by different test strategies. We conclude that, in the data-rich domain of fragrance ingredients, sensitization risk assessment without animal testing is possible in most cases by these integrated approaches.

Keywords: Skin sensitization, risk assessment, alternative methods, QRA, point of departure

Development and validation of computational models for predicting oxidative stress responses using comprehensive series of drug-like compounds (#231)

M. Pastor1, G. J. P. Westen2, J. C. Gómez-Tamayo1, B. E. Lenselink2, C. - C. Lam2, B. Water2, U. Norinder3, S. Manganelli4, D. Gadaleta4, A. Roncaglioni4

1 University Pompeu Fabra, DCEXS, Barcelona, Spain
2 Leiden University, Drug Discovery and Safety, Leiden, Netherlands
3 Swetox, Karolinska Institutet, Unit of Toxicological Sciences, Södertälje, Sweden
4 Istituto di Richerca Farmacologiche, Milan, Italy

Recent advances in machine-learning methods, including deep learning, are improving computational methods used for predicting biological endpoints. In this context, a relevant question is how far we are in producing realistic predictions of toxicological endpoints using in silico methods only.

With this aim, in the EU-ToxRisk project ( we developed machine learning models from a large series of over 2000 drug-like compounds, providing a comprehensive coverage of the druggable chemical space. For all these compounds, oxidative stress responses were measured using a high-throughput platform making use of GFP reporter system. The data was analyzed using a wide variety of state-of-the-art machine learning methods and molecular descriptors, generating a collection of more than 20 models. We systematically evaluated the predictive performance of the models, both individually and combined using diverse strategies. Furthermore, the models are being validated using a prospective approach by predicting the properties of a new series of compounds carefully selected to represent future query compounds, extracted from diverse regions of the chemical space of interest.

In this work we present the results of this analysis, including the quality of the predictions provided by the models (either as single models or combined), the uncertainty of the predictions and discus the usability of these approaches within an integrated risk assessment approach.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 681002.

Keywords: Machine-learning, Computational toxicology, In silico methods, Consensus models, Oxidative stress

Persistent organic pollutants, pre-pregnancy use of combined oral contraceptive and time-to-pregnancy in SELMA cohort (#255)

R. D. Björvang1, 2, C. Gennings3, P. - I. Lin4, G. Hussein5, 6, H. Kiviranta7, P. Rantakokko7, P. Ruokojärvi7, P. Damdimopoulou1, 2, C. - G. Bornehag3, 4

1 Karolinska Institutet, Department of Clinical Science, Intervention and Technology, Stockholm, Sweden
2 Swetox, Karolinska Institutet, Unit of Toxicological Sciences, Södertälje, Sweden
3 Mount Sinai, Department of Environmental Medicine and Public Health, New York, United States of America
4 Karlstad University, Department of Health Sciences, Karlstad, Sweden
5 Karlstad Central Hospital, Department of Obstetrics and Gynecology, Karlstad, Sweden
6 Sahlgrenska University Hospital, Department of Obstetrics and Gynecology, Gothenburg, Sweden
7 National Institute for Health and Welfare, Department of Health Security, Kuopio, Finland

Prior evidence has suggested that persistent organic pollutants (POPs) have adverse effects on reproductive function. The use of combined oral contraceptives (COC) and its association with return of fertility remains controversial. The goal of this study is to investigate the association between exposure to POPs and fecundity measured as time-to-pregnancy (TTP) according to the use of COC. Using the SELMA (Swedish Environmental Longitudinal Mother and Child, Allergy and Asthma) study, we have identified 818 pregnant women aged 17-43 years (mean 29 years) and asked how long they tried to get pregnant and what was their most recently used contraception method. Concentrations of 22 POPs were analyzed in the blood samples collected during their first prenatal visit, median 10 weeks age of gestation. To perform the subgroup analysis, we stratified the sample based on two factors, namely age (with 29 years of age as the cutoff) and use of COC as the most recent pre-pregnancy contraception method.  To study the chemicals individually, we used the Cox regression model to estimate fecundability ratios (FRs) and 95% CI. Weighted quantile sum (WQS) regression was done to investigate the chemicals as a mixture where bad actors (most influential compounds) were identified above the 7.6% threshold. The models were adjusted for parity, regularity of menses, maternal BMI and smoking status, and stratified as described above. In women aged 29-43 years who did not use COC as their most recent pre-pregnancy contraceptive, both the individual chemicals as well as a mixture were associated with lower fecundity. Specifically, the fourth quartiles of PCB 187, PCB 183 and PCB 180 have 34%, 31%, and 31% significantly longer TTP compared to their first quartiles, respectively. Regarding mixtures, there is a 7% increase in TTP for every quartile increase in the WQS index and the bad actors identified in this mixture were PCB 187, PCB 156, PCB 183 and transnonachlor. In conclusion, serum concentrations of POPs were significantly associated with lower fecundity, particularly in older women not using COC as their most recent pre-pregnancy contraceptive. Our findings suggest that pre-pregnancy use of oral contraceptive may modify the link between POPs and fecundity. We are currently studying whether the well-known impact of COCs on ATP-binding cassette (ABC) transporters could play a role in this phenomenon.

Keywords: time-to-pregnancy, combined oral contraceptives, persistent organic pollutants, fertility, environmental chemicals

In utero exposure to cigarette smoke and NAFLD pathways: sex and age specific effects in the human fetus (#494)

C. Talia1, P. Filis1, U. Soffientini5, B. Lucendo-Villarin2, A. Douglas1, D. Hay2, S. Shaw1, J. P. Iredale4, M. J. Swortwood3, M. A. Huestis3, M. Bellingham5, L. Connolly6, P. J. O’Shaughnessy5, P. A. Fowler1

1 University of Aberdeen, Institute of Medical Sciences, Aberdeen, United Kingdom
2 University of Edinburgh, Medical Research Council Centre for Regenerative Medicine, Edinburgh, United Kingdom
3 National Institutes of Health, Chemistry and Drug Metabolism, Intramural Research ProgramNational Institute on Drug Abuse, Baltimore, United States of America
4 University of Bristol, Senate House, Bristol, United Kingdom
5 University of Glasgow, Institute of Biodiversity, Animal Health and Comparative Medicine, Glasgow, United Kingdom
6 Queen's University Belfast, Institute for Global Food Security, Belfast, United Kingdom

Background: Non-alcoholic fatty liver disease (NAFLD) is the major cause of chronic liver disease worldwide. In utero exposure to endocrine disruptors might cause obesogenic effects and the reprogramming of hepatic set points, influencing the initiation of the disease later in life. Maternal smoking, a unique model of real early-life exposure, is associated with increased risk of obesity and metabolic syndrome in offspring. We investigated effects of maternal smoking on human fetal liver pathways related to NAFLD pathogenesis.


Methods: 80 human fetal livers from elective terminations of normal pregnancies (12-19 gestation week), were collected (NHS Grampian Research Ethics Committees, REC 04/S0802/21). RNA was extracted and the Illumina NextSeq platform produced 76 bp single end RNA sequencing reads. Reads were quality controlled, aligned to the human reference genome and quantified at gene regions. Significant differentially expressed genes were identified using a generalised linear model with a three-way interaction model between fetal sex, fetal age and maternal smoking status. Ingenuity Pathway Analysis (IPA) related the RNAseq data to biological functions and canonical pathways.


Results: Liver inflammation and steatosis were top-ranked pathways. In older smoke-exposed males, proinflammatory genes, CXCL8, JUN and SOCS3, were significantly upregulated. In males, maternal smoking abolished the developmental decrease of PTPN1, a critical negative regulator of insulin signalling, ACACA, a rate-limiting fatty acid synthesis enzyme, and SREBF2, a regulator of cholesterol synthesis. CPT1A, key gene of β-oxidation, increased throughout gestation in females only. PPARγ and FOXO1, transcription factors regulating lipid and glucose homeostasis, were significantly induced by smoking. These initial findings are being followed up in a larger study, combined with liver lipid analysis.

Conclusions: In utero exposure to smoking resulted in significant sex- and age-specific changes in multiple NAFLD-associated pathways. In smoke-exposed fetuses, the dysregulation of lipid homeostasis may lead to increased lipid storage and consequently development of steatosis. In particular, male smoke-exposed fetuses were the most affected, with proinflammatory and insulin-resistance reprogramming.

Keywords: human fetal liver, endocrine disruption, prenatal programming, Non-alcoholic fatty liver disease, maternal smoking

In vitro analysis of liver steatosis using adverse outcome pathways: a case study with cyproconazole (#657)

C. Luckert1, A. Braeuning1, G. de Sousa3, S. Durinck4, E. S. Katsanou5, P. Konstantinidou5, K. Machera5, E. S. Milani6, A. Peijnenburg2, R. Rahmani3, A. Rajkovic4, A. Spyropoulou5, M. Stamou6, G. Stoopen2, S. J. Sturla6, N. Zucchini-Pascal3, A. Lampen1

1 German Federal Institute for Risk Assessment, Food Safety, Berlin, Germany
2 RIKILT Wageningen University & Research, Wageningen, Netherlands
3 National Institute for Agricultural Research, TOXALIM Research Centre in Food Toxicology, Toulouse, France
4 Ghent University, Ghent, Belgium
5 Benaki Phytopathological Institute, Athens, Greece
6 ETH Zurich, Department of Health Sciences and Technology, Zurich, Switzerland

Adverse outcome pathways (AOPs) are a conceptual framework to describe causal relationships between molecular disturbance and adverse cellular effects. They are becoming increasingly important to link mechanistic toxicological knowledge from in vitro studies to in vivo data from regulatory toxicity studies. In this work, a bioassay toolbox was developed to assess central key events in the recently proposed AOP for chemically induced liver steatosis. The in vitro assays aim to measure nuclear receptor activation, gene and protein expression, lipid accumulation, mitochondrial respiration, and fattening of liver cells. Evaluation of assays was performed in human HepaRG hepatocarcinoma cells exposed to the model compound cyproconazole, a fungicide inducing steatosis in rodents. Concerning the molecular initiating event of nuclear receptor activation in the steatosis AOP, Cyproconazole dose-dependently activated RARα and PXR. Moreover, cyproconazole induced accumulation of triglycerides, provoked a disruption of mitochondrial functions and led to the formation of fatty liver cells as described in the AOP. However, a lacking mechanistic link between nuclear receptor activation and downstream events was revealed as gene and protein expression analysis failed to detect substantial expression changes as proposed in the AOP. Overall, cyproconazole was successfully shown to induce steatosis in human cells in vitro. This demonstrates the suitability of the developed bioassay toolbox in evaluating the perturbation of molecular and functional key events in the AOP which might eventually result in an adverse outcome. An AOP-driven in vitro testing strategy as demonstrated here thus might help to further improve existing AOPs, to elucidate molecular mechanisms of toxicity in human cell systems in vitro, and to assist risk assessment for individual compounds and their mixtures.This might be included in future toxicological testing strategies not only for individual compounds, but also for mixtures, a strategy which is, for example, pursued within the European project “European Test and Risk Assessment Strategies for Mixtures” (EuroMix;

Keywords: AOP, in vitro, bioassay toolbox, HepaRG, cyproconazole, EuroMix

Limitations of the T25 method for the assessment of carcinogenic potency of inorganic inhalation carcinogens (#660)

R. Danzeisen1, V. Verougstraete2

1 Cobalt Institute, Guildford, United Kingdom
2 Eurometaux, Brussels, Belgium

The “T25 method” is a tool used in chemicals classification (EU Classification Labelling and Packaging; CLP) to decide on the carcinogenic potency of substances. The method is based on a publication by Dybing et al. (1997), and groups carcinogens into “potency classes” by comparison to a database with organic substances that are known systemic carcinogens by oral exposure. Based on the potency class a cut-off is identified that will be used for the classification of mixtures including the classified substance, or when it is contained as an impurity. The purpose of this work was to evaluate whether this method is suitable for inorganic inhalation carcinogens.

All existing NTP inhalation cancer studies carried out since 1993 with inorganic substances were reviewed and the most sensitive species was selected. The T25 metric was calculated by determination of the lifetime exposure level resulting in a 25% increase in cancer above background cancer incidence and converting this exposure level into the “systemic dose” following current EU CLP guidance. The carcinogenic potency was assigned based on EU CLP guidance.

More than 80% of inorganic inhalation carcinogens are predicted by the T25 method to be of “high potency”, based on a very low T25 value. This includes inorganics of low toxicity, “inert dusts” and substances with negative epidemiological cancer data.

Reasons for the extreme conservatism of T25 when applied to inorganic inhalation carcinogens were explored, and may include

  • (i) the assumption of a linear dose response by the T25 method
  • (ii) lack of a possibility to accommodate thresholds of effect or information at low doses/exposures
  • (iii) application of generic conversion factors when translating inhalation exposure concentrations into systemic dose
  • (iv) requirement to compare all substances and all exposure routes against a database limited to organic systemic carcinogens with ranges selected arbitrarily in order to place the majority of the carcinogens in the original database into the medium potency range.

In view of the important consequences on chemicals management, a possible refinement for inorganics is proposed.


Keywords: risk assessment, dose response, hazard assessment, metals