Current methods in risk assessment of genotoxic chemicals

Food and Chemical Toxicology xxx (2016) 1e9

Contents lists available at ScienceDirect

Food and Chemical Toxicology journal homepage: www.elsevier.com/locate/foodchemtox

Current methods in risk assessment of genotoxic chemicals Alexander Cartus*, Dieter Schrenk University of Kaiserslautern, Food Chemistry and Toxicology, Erwin-Schroedinger-Strasse 52, 67663, Kaiserslautern, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 July 2016 Received in revised form 6 September 2016 Accepted 8 September 2016 Available online xxx

Chemical contaminants and residues are undesired chemicals occurring in consumer products such as food and drugs, at the workplace and in the environment, i.e. in air, soil and water. These compounds can be detected even at very low concentrations and lead frequently to considerable concerns among consumers and in the media. Thus it is a major challenge for modern toxicology to provide transparent and versatile tools for the risk assessment of such compounds in particular with respect to human health. Well-known examples of toxic contaminants are dioxins or mercury (in the environment), mycotoxins (from infections by molds) or acrylamide (from thermal treatment of food). The process of toxicological risk assessment of such chemicals is based on i) the knowledge of their contents in food, air, water etc., ii) the routes and extent of exposure of humans, iii) the toxicological properties of the compound, and, iv) its mode(s) of action. In this process quantitative dose-response relationships, usually in experimental animals, are of outstanding importance. For a successful risk assessment, in particular of genotoxic chemicals, several conditions and models such as the Margin of Exposure (MoE) approach or the Threshold of Toxicological Concern (TTC) concept exist, which will be discussed. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Risk assessment Genotoxicity testing Contaminants Residues Carcinogenicity Genotoxicity

1. Introduction Unwanted compounds or impurities of toxicological concern in consumer products such as food or pharmaceutical drugs can be divided into contaminants and residues (Codex Alimentarius Commission, 2015). Contaminants are compounds which are found in environmental samples and food etc. Per definition, contaminants were not released or added intentionally to consumer products. Their occurrence usually is the consequence of insufficient prevention and/or cleaning precautions related to thermal or other chemical processes and/or of other inadequate production, handling or manufacturing of chemicals or during processes. Examples in food and feed are dioxins, mycotoxins and pyrrolizidine alkaloids (most frequently due to co-harvesting) or compounds formed during the production process such as acrylamide and ethyl carbamate. An exception, i.e., an intended introduction of a contaminant may occur for criminal purposes such as the illegal ‘disposal’ of contaminated materials in the feed chain of livestock. Examples are aniline (Spanish toxic oil syndrome) or polychlorinated biphenyls (PCBs) and dioxins which can enter the food chain via

* Corresponding author. E-mail address: [email protected] (A. Cartus).

contaminated oil (Diggle, 2001; Bernard et al., 2002). Besides contaminants there are residues present in consumer products. These are derived from the intentional use during the production process, mainly during plant protection (e.g. from the use of pesticides) or from the use of veterinary drugs etc. during agricultural production of animal products such as meat, fish, eggs or milk. Such compounds may also occur as contaminants as long as they originate from intentional use for such purposes but contaminate secondary targets such as soils or surface water (e.g. from the use of contaminated manure or from drift during plant protection measures). However, this rigid classification cannot be applied strictly to all chemicals of toxicological concern in consumer products. Examples are plant-derived compounds such as alkenylbenzenes (e.g. estragole, methyleugenol) in herbs and spices or coumarin in cinnamon which are usually not regarded as contaminants because they are naturally occurring constituents in certain plants. Furthermore, ‘genotoxic impurities’ in pharmaceutical drugs may be introduced or arise during the synthesis of active ingredients and can be considered as contaminants (reaction side products) or residues (residual reagents such as heavy-metal catalysts, solvents or starting material; Szekely et al., 2015). The distinction between contaminants and residues has extensive consequences for the risk assessment of these classes. Residues in a consumer product result from their intended use. Therefore, in

http://dx.doi.org/10.1016/j.fct.2016.09.012 0278-6915/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

2

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

most cases such compounds have to go through a thorough legislative approval process where a risk assessment for the compound is made with regard for the purpose of use. For example, an active substance in a plant protection product has to be evaluated and approved (“positive list”). This requires a comprehensive toxicological characterization of the compound to make sure that no (or a justifiable) risk will emanate from the use of the compound (in the final product/formulation). The manufacturer of such a chemical is responsible to provide sufficient toxicological data as a prerequisite for the legal authorization. In contrast, contaminants enter consumer products unintentionally. Therefore, the structural identification of contaminants, as well as the synthesis or production of sufficient amounts of “new contaminants” needed for toxicological tests can be an obstacle for the further toxicological characterization in many cases. The lack of detailed chemical and toxicological information of a certain contaminant concerns all steps of risk assessment and is hence a major challenge illustrating the need for alternative methodologies beside the classical approaches. This paper will focus on food-borne toxicants with special emphasis but not exclusively on non-genotoxic and genotoxic carcinogens. Certain points concerning genotoxicity test methodologies, regulations and implications for the risk assessment are presented in more detail in the other articles being part of this Special Issue. 2. Utilization of data in toxicological risk assessment Toxicological risk assessment of chemicals typically consists of four steps: 1. Hazard identification (i.e. determination of substances that may have inherently adverse effects under certain conditions of exposure). 2. Hazard characterization (i.e. the qualitative and desirably quantitative description of the nature of the hazard, such as toxicokinetics, mechanisms of action and dose-response relationships). 3. Exposure assessment (addressing the quantitative question of how much of a certain substance a defined population will be exposed to) and 4. Risk characterization (consolidation of evidence, reasoning, and conclusions gained and collected in steps 1e3. and the estimation of the probability of the occurrence of an adverse effect in a certain population, taking several uncertainties into account).

2.1. Hazard identification and characterization Toxicological risk assessment of a compound starts with a detailed description of its chemical structure and the physicochemical properties. Furthermore, the pathways of formation or synthesis and the methods for quantitative analysis of the compound should be documented and discussed. In the next step, the toxicologically relevant facts about the behavior and effects of the compound in biological systems (in vivo and/or in vitro) of interest are collected. Studies using in vivomodels are also designed to obtain information about the bio-, pharmaco- or toxicokinetic of the compound. This includes data related to the absorption from the gastro-intestinal tract, lung, skin etc., metabolism, distribution in the organism and elimination. Of course, the intact organism is the best model to gain such data. However, in vitro models at various levels exist, which may allow estimates for biokinetic parameters describing the aforementioned

processes. The next step in risk assessment is the thorough investigation of the adverse effects elicited by the compound. Biological systems to study many toxicological endpoints mainly comprise higher organisms such as experimental animals (most frequently mice and rats) and organs, cells, subcellular fractions or purified target molecules (e.g. enzymes) isolated thereof or even more complex ‘organ- or human-on-a-chip’ approaches (An et al., 2015), often in combination with in silico-calculations, for example via physiologically based biokinetic modeling (Sung et al., 2014; Abaci and Shuler, 2015; Lin et al., 2016). Since toxicity studies in humans cannot be carried out for both ethical and legal reasons (with the exception of testing mild effects on the skin etc.), also human primary cells, subcellular organelles (e.g. mitochondria or microsomes), enzymes, receptors etc. are studied. Furthermore, studies in established human or other mammalian, so-called permanent cell lines are common. In vitro investigations can play an important role in defining a mode of action (MoA) of a compound and are often initiated by findings from animal experiments. Frequently, primary cells isolated from animal or human tissues are used for this purpose. These show the advantage of being relatively similar to the cells in situ of the organism reflecting the responses of the latter towards chemical challenges. Unfortunately, such cells loose many of their cell- or tissue-specific properties after having been kept in culture for hours, days or weeks. Thus, animals must be sacrificed regularly in order to isolate fresh primary cells. Cell lines are widely used and a well alternative to in vivo experiments for several investigations. Once established in a laboratory, their availability is nearly unlimited, handling is comparatively easy and cheap. However, they often have very few specific properties in common with the parent cells in situ, which make them less useful if the knowledge about the toxicological properties of a compound is still low. Moreover, cell lines change their properties further once being transferred and kept over several passages. This leads to the situation that ‘the same’ cell line can differ when propagated over several cell generations in different laboratories which may limit their usefulness for toxicological research sometimes. However, progress has been made in this regard using 2D- or 3D (co-)culture models for many organs such as liver or skin and for certain types of
pe
e et al., 2014; Nath and Devi, 2016). Taken together cancer (Ale the predictive power of such in vitro methods is frequently overestimated with respect to adverse effects in humans and animals. However, such methods can be highly valuable, if a well-defined mechanistic hypothesis needs to be tested. For this purpose, the in vitro models should be as well characterized as possible. Examples are agonistic or antagonistic effects at certain receptors, the induction of apoptosis, cell division etc. Again, one of the most reliable sources for toxicological data are animal experiments or, if available, observations in humans. Only these systems can provide information of reactions in a system as complex as the intact organism. Animal experiments are mostly carried out as acute, sub-acute or chronic feeding or treatment studies where the test compound is applied via the relevant route, for example oral (i.e. in the feed, the drinking water or per gavage) or per inhalation. Other application routes, for example (sub) cutaneous or intravenous administration are considered less relevant unless a compound is suspected to enter the body via such a route which may be relevant for a contaminant in a medical preparation intended for injection or dermal application etc. Particularly valuable if not compulsory for quantitative risk assessment is the use of various dose levels. Among those doses, at least one should elicit clear effects whereas other should be without (adverse) effects (Bucher, 2002; Rhomberg et al., 2007). Furthermore, a statistically sufficient number of animals should be used. All relevant findings and observations must be documented

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

and/or published and the report/paper should contain a critical appraisal of putative deficits or sources of bias or error. Findings and observations in a control group of sufficient size is a particularly critical feature of a useful study. Finally, the ‘adverse’ effect should be of clear clinical or physiological relevance and should be biologically plausible and statistically significant. Furthermore, the effect should increase with dose in a monotonous manner. Nonmonotonous dose-response relationships, likely to exist in toxicology, for example in the case of endocrine disruptors (Lagarde et al., 2015), require particularly careful testing since they appear to be the exception rather than the rule. These conditions must be fulfilled altogether in a study providing a convincing basis for a risk assessment. However, in too many toxicological studies, biological variability and background alterations in the control cohort such as age-dependent changes in cancer incidences or degenerative events, as well as the optimal use of historical control data, are not sufficiently taken into account (Maronpot et al., 2016; Hayashi et al., 2011). Ideally, mechanistic findings which allow an understanding of the mode(s) of action (MoA; on a functional cellular level) and mechanism(s) of action (on a molecular level) of the chemical are available. Alternatively a number of essential key events are identified allowing the correlation to a known adverse outcome pathway (AOP) which is thus less dependent on knowledge about certain apical toxicological endpoints. In these cases, the scientific confidence in the risk assessment can be extraordinarily high, in particular if related compounds have already shown similar effects. In addition the understanding of a MoA is indispensable for evidence-based decisions about the relevance of results from animal experiments for the human situation. Such decisions are not trivial since the widely used rodent species are quite distinct form humans with respect to many of their physiological functions. Limitations are, for example, differences in toxicokinetic parameters as well as differences or the lack of enzyme activities in one of the species. In many instances, chemical substances interact with various targets in the organism and can thus elicit a whole pattern of effects or symptoms, which are not necessarily based on a common MoA. A good example is cadmium, which can affect both bone mineral structure and kidney function (Rani et al., 2014; Burm et al., 2015). Both effects are depending on different types of interactions with the targets. In such cases, the more sensitive endpoint may be chosen as a basis for risk assessment since protective or mitigation/limitation measures then would protect both the more and the less sensitive target. In any case, a detailed consideration on the nature (MoA and mechanism of action) and sensitivity (dose-response-relationship) of the various endpoints is needed. More complex is this decision if carcinogenicity is one of the critical endpoints. Under such conditions, more detailed information on the MoA, i.e. genotoxic vs. non-genotoxic, is compulsory. The other way around, genotoxicity is not only associated with cancer, but also with degenerative conditions such as accelerated aging, immune dysfunction, cardiovascular and neurodegenerative diseases (in case of accumulation of DNA damage in somatic cells) and with spontaneous abortions, infertility or heritable damage in the offspring and/or subsequent generations resulting in genetic diseases (in case of DNA damage in germ cells). Consequently, genotoxicity data are requested in almost all regulatory settings. 2.2. Exposure assessment Investigation of the levels of compounds of toxicological concern in various environmental media, drugs, and food etc. represents another cornerstone in risk assessment. Finally,

3

exposure should be estimated preferentially using appropriate models for routes such as via ingestion of food, use of drugs, inhalation, direct skin contact etc. Since the focus of this paper is on food-related contaminants and residues, the oral route will be the most important rather than inhalation or skin absorption. Whereas human risk assessment was long time hazard-driven, a paradigm change to a more and more exposure-driven process can be observed (EC, 2012) which bears several advantages (Burden et al., 2015). Examples of rather exposure-driven approaches are the development of the AOP concept as it starts with the exposure (Vinken, 2013) or the below mentioned Threshold of Toxicological Concern (TTC) concept in which more emphasis is put on the exposure part. It is evident that the precision of such estimates strongly depends on the numbers and representativity of the analyzed samples and the identification and analysis of biomarkers for exposure. Likewise, exposure via food depends on a thorough knowledge on food consumption for various food categories, e.g. among European countries. Chronic and acute food consumption statistics for several populations are available, for example freely provided by the EFSA Comprehensive European Food Consumption Database in Exposure Assessment (EFSA, 2016). Average levels of the compound in such categories can be used for exposure assessment. A detailed analysis of subgroups allows a fine-tuning of such estimates possibly identifying groups of consumers which are at particular risk. For various sub-populations probabilistic approaches have been established by using distributions instead of fixed values for quantifying sources of variability and uncertainty affecting exposure. Thus they provide a basis to model and estimate average or high-level exposure. As an example, guidance on the use of probabilistic methodology for modeling dietary exposure to pesticide residues is given by EFSA (2012). 3. Characterization of a compound as genotoxic and carcinogenic based on results from in vivo and in vitro studies For genotoxicity testing, a battery of in vitro and/or in vivo tests is needed to cover all endpoints (i.e. gene mutations and structural and numerical chromosome damage). In vitro methods are particularly suitable to investigate the possible genotoxicity of a compound (see also section 2: Hazard identification and characterization). For this purpose, several in vitro tests in bacteria, most notably the Ames test, or in mammalian cells are widely used. These in vitro genotoxicity tests often require the presence of an enzyme or a protein mixture along with several co-factors which are not expressed or not functional in bacteria or certain mammalian cell lines. Such exogenous metabolic activation systems can be, for example, liver homogenates or other liver preparations (e.g. S9-mix) together with NADPH or an NADPHgenerating system frequently used in the Ames test (see also Nesslany, 2016). Table 1 summarizes selected in vivo and in vitro tests to assess possible genotoxic effects of a compound. A positive outcome is due to an interaction (e.g. a covalent binding) of the compound or (a) metabolite(s) with DNA (direct interaction) or to a DNA-processing target (e.g. a microtubular protein) as an indirect interaction. As a consequence, a persistent change in the DNA base sequence and/or chromosomal structure or even a complete loss of parts of the cellular genome can occur. Such changes are often linked to an increased tumor incidence if they would have occurred in an intact multicellular organism if special genes/gene areas are affected and if certain other circumstances in the multistep process of cancer development come together (i.e. inefficient DNA repair, suppression of apoptosis etc.). This is also one of the reasons why a positive and reliable prediction on the carcinogenicity of a compound in vivo cannot be derived from a

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

4

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

Table 1 Tests for the detection of genotoxicity in vitro and in vivo. End point

Test

in vitro test methods Mutagenicity (reverse mutation) Ames test, Ames fluctuation test Mutagenicity (forward mutation) Mutagenicity (forward mutation)/Chromosomal damage Chromosomal damage Chromosomal damage DNA strand breaks Induction of DNA repair DNA adducts in vivo test methods DNA strand breaks Induction of DNA repair DNA adducts Mutagenicity Chromosomal damage Chromosomal damage a

Hprt test Thymidine kinase-/Mouse lymphoma assay

Species

Referencesa

Bacteria (Salmonella thyphimurium, Escherichia coli) Mammalian cell lines TK6 human lymphoblastoid cell line; L5178Y mouse lymphoma cell line

OECD TG 471; Ames et al., 1973; Mortelmans and Zeiger, 2000; Reifferscheid et al., 2012 OECD TG 476; Albertini, 2001 OECD TG 490

Chromosome aberration test in vitro Mammalian cell lines Micronucleus test in vitro Human and mammalian primary lymphocytes and cell lines Comet assay Cells and cell lines Unscheduled DNA synthesis in vitro Cells and cell lines 32 P Postlabeling, mass spectrometry Cells and cell lines and Immunoassays

OECD TG 473 OECD TG 487, Araldi et al., 2015

Comet assay Unscheduled DNA synthesis in vivo 32 P Postlabeling, mass spectrometry and Immunoassays Transgenic rodent somatic and germ cell gene mutation assays Micronucleus test in vivo Chromosome aberration test in vivo

OECD TG 489 OECD TG 486 Keith and Dirheimer, 1995; Balbo et al., 2014; Farmer and Singh, 2008; Santella, 1999 OECD TG 488

Mammalian tissues Mammalian liver cells Mammalian tissue and peripheral blood cells Transgenic rats or mice

Araldi et al., 2015 Madle et al., 1994 Keith and Dirheimer, 1995; Balbo et al., 2014; Farmer and Singh, 2008; Santella, 1999

Mammal erythrocytes/blood cells OECD TG 474; Araldi et al., 2015 Mammalian bone marrow and mammalian OECD TG 475; OECD TG 483 spermatogonial cells

OECD Guidelines for the Testing of Chemicals are available online at http://www.oecd-ilibrary.org/.

positive in vitro outcome only. Additionally, in vitro tests usually do not completely reflect parameters of absorption, distribution and elimination (toxicokinetics) or of special metabolic pathways needed for the activation of the compound. Furthermore, physical factors such as oxygen pressure, light exposure etc. can differ between a cell incubator and an intact organ. In some cases, the compound has already been tested in a chronic animal experiment, for example in a carcinogenicity study or a combined chronic toxicity/carcinogenicity study (OECDGuidelines 451 and 453). These tests primarily use rats and/or mice to detect neoplastic effects and the general toxicity, respectively, over the life-span of the animals under chronic exposure conditions. If a compound turned out to be carcinogenic, in vitro tests can provide additional crucial information on the MoA, most notably to clarify the question if the carcinogenicity is a result of a genotoxic MoA. In some cases (see above), the in vitro genotoxicity assay can be false negative, i.e. cannot reflect the complex in vivo situation for example due to the lack of enzymes or co-factors present in the test system. In some cases, however, the in vitro genotoxicity assay can also appear false positive, i.e., the compound is not genotoxic in vivo but exerts a carcinogenic effect through a non-genotoxic MoA (Fowler et al., 2012; Nesslany, 2016). In these cases, in vivo genotoxicity tests are needed to obtain more information. On the other hand, if the compound was not carcinogenic, it can be classified as not genotoxic in vivo. Further information on in vitro genotoxicity testing, e.g. test batteries and regulatory requirements is given by Corvi and Madia (2016).

Point of Departure (POD) for the further risk assessment. Most simply, a No Observed Adverse Effect Level (NOAEL), i.e., the highest dose level which did not significantly cause an adverse effect (Fig. 1), can be used for the assessment of non-genotoxic compounds but is not considered appropriate for the assessment of substances that are both genotoxic and carcinogenic. Modeled PODs are calculated by fitting various mathematical functions to the dose-response data. The modeled functions comprise statistically defined confidence rooms, so-called confidence intervals. Having defined a critical (benchmark) level of effect in advance, the modeled function allows estimating the dose levels which would elicit this degree of effect (e.g. a 10% decrease of L-thyroxine in peripheral blood or a 10% increase in cancer incidences). This level is then termed Benchmark Dose with 10% effect (BMD10). The lower level of the corresponding confidence interval (e.g. 95% confidence) allows the derivation of a Lower Confidence Level of a Benchmark Dose with 10% effect (BMDL10; Davis et al., 2011) what is depicted in Fig. 1. Also BMDLs at other effect levels such as a BMDL01 or a

4. Risk characterization and recommendations for healthbased guidance values 4.1. Dose-response relationship and point of departure After having identified critical endpoints in animal experiments among the carcinogenic and non-carcinogenic effects (e.g. neurotoxicity, immunotoxicity, endocrine disruption or reproductive toxicity), the dose-response data can be used for the creation of quantitative dose-response relationships and the definition of a

Fig. 1. Modeling of experimental data to derive the Lower Confidence Level of a Benchmark Dose with 10% effect (BMDL10, black) and data used for linear intra- or extrapolation (grey). Note that the effect in case of genotoxic carcinogens may be very low (or even not detectable/measurable) but not necessarily zero at low doses.

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

BMDL05 are frequently calculated (e.g. Smith et al., 2010). For the BMD modeling, free in silico software is available such as BMDS (US EPA) or PROAST (RIVM). The BMD approach is independent from the exact dose levels used in the experiments whereas the NOAEL bears the disadvantage that it depends strictly on the used dose levels (and sample size). The BMD approach, however, requires that the dose levels used cover the relevant dose range (including ineffective and effective doses, i.e. representing a NOAEL, a Lowest Observed Adverse Effect Level, LOAEL, and increasing doses with concomitant increasing effects) and allow a mathematical modeling, i.e., a certain number of data points are required and need to be in a reasonable order/shape to allow a mathematical fit. Either the BMDL10, the T25 (i.e. the dose that causes 25% tumor incidence in experimental animals by linear interpolation, Sanner et al., 2001), TD50 (i.e. the daily dose rate required to halve the probability of remaining tumor-free animals at the end of life span; Peto et al., 1984) or a signal-to-noise crossover dose (defined as the dose where the additional risk is equal to the “background noise”; Sand et al., 2011) may be used as POD for genotoxic carcinogens, i.e. as starting point for the further risk assessment. Table 2 gives a simple example on the calculation of a T25 by linear intrapolation. Therefore, usually the lowest tumorigenic dose that results in a statistically significant increase of the incidence at the most sensitive tumor site (benign and malignant tumors) from a chronic oral toxicity or carcinogenicity study (which has to be also relevant to humans) is used. However, this procedure requires a clear picture of the critical endpoint(s), the MoA, and the relevant exposure(s). 4.2. Non-genotoxic compounds In case of a non-genotoxic endpoint, a threshold, i.e. a dose which does not result in an adverse effect, is widely assumed in toxicology. As a consequence, no adverse effects can be expected below this dose. Since the data are usually derived from animal experiments and not from humans an uncertainty factor (UF; also called ‘safety factor’, dimensionless) is applied. Because individuals among the human population can differ widely in the susceptibility towards a chemical (not to be confused with allergic or other immune reactions!), an additional UF is used. In many instances, this leads to a composite UF, e.g. of 10
10 ¼ 100 (10 for the extrapolation from animals to humans and 10 for the intra-species differences). Division of the POD by the UF results in a so-called healthbased guidance value, usually a Tolerable Daily Intake (TDI, used for contaminants), Acceptable Daily Intake (ADI, used for food additives and residues) or Reference Dose (RfD) which describes the dose of a chemical tolerable for humans on a daily basis and standardized to the body weight (e.g. as mg/kg b.w. per day).

TDI or ADI ¼

NOAEL UF

The most often used POD to derive an ADI or a TDI is the lowest NOAEL obtained in all experiments and for all investigated endpoints. If a NOAEL could not be derived, i.e., all dose groups show effects, the LOAEL may be used instead using an additional UF (often 3e10) but this should not be regarded to be a LOAEL-toNOAEL extrapolation. The other way around, dividing the NOAEL by the exposure, the result is a so-called Margin of Safety (MoS), e.g. to compare different compounds of concern. Since both NOAEL and exposure should have the dimension of a dose, the resulting MoS is dimensionless.

MoS ¼

NOAEL Exposure

In cases where compounds have a tendency to accumulate in certain organs or compartments of the body, such as cadmium, dioxins or ochratoxin A, longer periods of consideration such as the Tolerable Weekly Intake (TWI) may be appropriate. Usually, TDI values are valid for all members of the population independent of gender, age, race etc. In some cases, however, the MoA points to specific vulnerable groups which need to be addressed separately. 4.3. Genotoxic compounds In case of good evidence for genotoxicity, however, it is commonly accepted that a threshold of effect cannot be assumed. Although a variety of in vivo and in vitro findings argue for the existence of such as ‘practical’ thresholds (Thomas et al., 2015; Fahrer et al., 2015; Wei et al., 2012; Fukushima et al., 2016), its existence cannot be considered as a general principle. The rational for this paradigm is the assumption that theoretically a single point mutation, i.e., as a consequence of an interaction of one molecule with the DNA, could result in a tumor cell. From the fact that our genome is targeted every day by a considerable amount of so-called ‘background factors’ (natural irradiation, naturally occurring radioactive isotopes) without developing a tumor, indicates that such an event is highly unlikely. Nevertheless, the genotoxic carcinogens widely considered as active in humans are mostly manmade and as such could theoretically be avoided. This fact encourages a reduction policy for those compounds such as the ALARA (As Low As Reasonably [or Technically] Achievable) principle. There, risk managers, manufacturers, regulators, and scientists try to come to solutions which are aimed at reducing the exposure as far as economically and technically possible. In the reality, it is sometimes not possible to reduce the exposure substantially without unacceptably affecting unavoidable production processes or technical procedures in everyday's life. At best gradual reductions towards a certain degree appear achievable such as in the case of acrylamide formation in fried potato products including home and restaurant cooking, or the formation of carcinogenic aromatic hydrocarbons during barbecuing. However political and

Table 2 Linear interpolation to calculate a T25 from a rodent study (fictitious values). Dose group

Tumor incidence (most sensitive site)

Control 0 mg/kg b.w per day Low dose 16 mg/kg b.w per day Mid dose 160 mg/kg b.w per day High dose 480 mg/kg b.w per day

0/50 animals ¼ 0%

a

3/50 animals ¼ 6% 17/50 animals ¼ 34%a

5

T25 ¼ 160
25% 34% ¼ 118 mg=kg b:w:per day

46/50 animals ¼ 92%a

Statistically significant.

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

6

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

economic factors are taken into account or may play a role when making administrative decisions. Here, communication and education play a central role. In the past, risk assessors tried to estimate additional cancer cases based on linear extrapolation of a dose-response relationship from tumor incidences in experimental animals. This procedure had the enormous disadvantage that the calculations were based on average exposures by the general population, i.e., did not consider extraordinarily exposed people with respect to a certain carcinogen. Furthermore, the stochastic nature of the risk estimate results in the fact that the putative cancer cases can never be identified as such. Thus, the concept can never be verified or even tested. The fact that it uses a linear extrapolation model over several orders of magnitude makes it extremely vulnerable towards apparently ‘slight’ deviations from linearity. Because of these drawbacks and limitations in risk communication the Margin of Exposure (MoE) concept has been propagated. It compares the actual average or high exposure in the human population of interest with the POD (e.g. T25, but most recommended BMDL10) from a cancer experiment in animals. Again, both POD and exposure should have the dimension of a dose (i.e. mg/kg b.w. per day), thus resulting in a dimensionless MoE.

MoE ¼

POD ðe:g: BMDL10 Þ Exposure

The European Food Safety Authority (EFSA) has applied this concept recently to the risk assessment of genotoxic carcinogens in food, i.e. to acrylamide in food (EFSA, 2015) and to pyrrolizidine alkaloids in honey (EFSA, 2011). A general opinion on the use of the MoE concept had suggested that a MoE of >10 000 should be considered as ‘of low concern’ while a MoE 10 000 should be considered as ‘of concern’, “if it is based on the BMDL10 from an animal carcinogenicity study, and taking into account overall uncertainties in the interpretation” (EFSA, 2012b). The number of 10 000 is derived by factors of 10 for i) inter-species extrapolation, ii) intra-species differences, iii) additional special variability which extends beyond the variability already obtained in the customarily applied factor 10 for the intra-species difference and concerns the individual cancer risk (depending, for instance, on DNA repair activity and cell cycle control)”, and iv) the fact that the BMDL is not an adequate substitute for a threshold value for tumor induction. A further additional factor of 2.5 is applied when using a T25 instead of a BMDL10 value. (BfR, 2005). The level of 10 000 thus has to be used as a ‘technical signal level’ which describes a rough classification of different degrees of concern. These degrees are meant to be translated into a lower or higher degree of urgency of measures to reduce exposure. Unfortunately, this approach has led to a number of misconceptions. The first misconception is, that a MoE >10 000 should be considered as ‘safe’. Following the paradigm of a lack of threshold for genotoxic carcinogens this assumption is fundamentally wrong. In contrast, a continuous dose-response relationship between the dose and the tumor outcome has to be expected. Nevertheless, it is obvious that higher dosages lead to a higher cancer risk, lower dosages to a lower one, while completely non-carcinogenic dose levels cannot be expected to exist. These facts represent a great challenge for risk managers and communicators since the public and the published opinion seem to expect that our environment and all products are absolutely safe. This expectation cannot be fulfilled. A second misconception prevails in the field of drug safety. There, the MoE approach is misused the other way around for the calculation of ‘acceptable’ levels, such as in the case of pyrrolizidine alkaloids in herbal medicines (Allgaier and Franz, 2015). In fact, it is possible to calculate a value by dividing the POD by 10 000 reaching

a level of low concern. However, such a procedure was not intended when the concept was designed because it may suggest to the public that levels below the ‘new TDI’ were ‘absolutely safe’, which is against the scientific paradigm of genotoxicity. In contrast, the MoE method is intended to create a level of concern which may force the involved parties to enhance their efforts to apply, e.g. the ALARA principle more effectively. Especially for regulatory purposes, a so-called ‘virtually safe dose’ may be derived by linear extrapolation of e.g. TD50 values down to an upper-bound limit of a lifetime cancer risk excess of one in a million which was used by FDA to set limits on the migration of chemicals from food contact materials to a value of 0.5 ppb in the foodstuff (Cheeseman et al., 1999). However, Cheeseman et al. identified several classes of chemicals such as aflatoxines, steroids or dioxins which were more potent, so that a limit of 0.5 ppb would result in a higher additional cancer rate than 1:106. Based on these findings, the concept of Threshold of Toxicological Concern (TTC) has been developed by Kroes et al. (2004). It was aimed for low levels of chemicals of known structure (and exposure) but lacking experimental data which were sufficient for a full quantitative risk assessment. For those compounds, a readacross approach makes use of toxicological data of chemicals bearing related sub-structures. Such so-called alerts may indicate possible genotoxicity and carcinogenicity and thus are termed genotoxicity/carcinogenicity alerts. Examples are an N-nitroso moiety or a hydrazine group within the molecule of interest. As a next step the quantitative dose-response relationships for the related carcinogenic molecules were taken together and the risk of tumor development of 1 in a million based on an average food consumption of an adult human individual was calculated. It turned out that most members of a 'structure-related group' exhibiting a genotoxic alert (certain special compounds excluded) exert a negligible risk when ingested at a dose level of 0.15 mg/day or below. Table 3 summarizes the TTC values derived by Kroes et al. (2004; 2007; see there for the TTC decision tree) and extended by EFSA (2012c). The first decisions are to evaluate if the compound is possibly genotoxic (threshold 0.15 mg/day), or if the compound is an organophosphate or a carbamate (threshold 18 mg/day), or if it belongs to a special group of chemicals (e.g. high potency carcinogens) for which the TTC approach is not applicable (see exclusion list in Table 3). Other compounds without a genotoxic alert will be classified according to their toxicity due to the rules of Cramer et al. (1978) in Cramer class I to III. Therefore, in silico tools have been developed such as the open source software Toxtree for hazard estimation (http://toxtree.sourceforge.net/) which provides beside Cramer classification many other endpoint predictions such as carcinogenicity, mutagenicity, skin sensitization, cytochrome P450metabolism or DNA- and protein binding properties. The TTC approach is frequently used for the risk assessment of food contact materials, flavoring substances, pesticide metabolites in groundwater (EFSA, 2012b) but is most widely used to assess genotoxic impurities in pharmaceuticals. Rarely, the TTC concept is used for food contaminants such as certain Alternaria toxins (EFSA, 2011b). However, many other applications seem possible (Szekely et al., 2015; SCCS, 2012; Blaauboer et al., 2016) but may need additional adjustments or refinements, for example for the application to substances in cosmetics since the TTC concept has been developed based on data of oral studies and for systemic toxicity whereas for cosmetics the dermal route (and possible local toxicity) is more relevant (Williams et al., 2016; Re et al., 2009; Safford, 2008). Again, misconceptions prevail among the applicants and users of the TTC concept. One is the misuse of this concept for chemically undefined mixtures in herbal medicines assuming that a worst-case scenario would be that the complete herbal extract is a compound with a genotoxicity/carcinogenicity alert. This is very

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

7

Table 3 Thresholds of toxicological concern derived by Kroes et al. (2004, 2007) and extended by EFSA (2012c). Compound class

Thresholda

Action/Result

with structural alert for genotoxicity, not belonging to another class

0.15 mg/person/day (0.0025 mg/kg b.w. per day) 1.5 mg/person/day (0.025 mg/kg b.w. per day) 90 mg/person/day (1.5 mg/kg b.w. per day) 540 mg/person/dayb (9 mg/ kg b.w. per day) 1800 mg/person/day (30 mg/ kg b.w. per day) 18 mg/person/day (0.3 mg/kg b.w. per day) not applicable

Negligible risk (low probability of a life-time cancer risk > 1:106) Substance would not be expected to be a safety concern Substance would not be expected to be a safety concern Substance would not be expected to be a safety concern Substance would not be expected to be a safety concern Substance would not be expected to be a safety concern Compound-specific data needed for risk assessment

all compounds without structural alert for genotoxicityc Cramer class III Cramer class II Cramer class I Organophosphates and carbamate substances with anti-cholinesterase activity Aflatoxin-like, azoxy- and N-nitroso compounds, benzidines, hydrazines; proteins; steroids; inorganic substances; metals and metal-containing compounds, substances that are known or predicted to bioaccumulate, i.e. polyhalogenateddibenzodioxins, -dibenzofurans, -biphenyls,; nanomaterials; radioactive substances; mixtures of substances containing unknown chemical structures a b c

If the exposure exceeds the TTC, compound-specific data are needed for risk assessment. EFSA (2012c) recommended for Cramer class II substances the same threshold as for Cramer class III substances, i.e. 90 mg/person/day. EFSA (2012c) did not integrate this step in their refined TTC approach since these compounds can be classified in one group below.

unrealistic and strongly overestimates any cancer risk. Secondly, manufacturers and regulators tend to use the TTC concept in parallel to a conventional risk assessment (see above), not at least to see which value is the ‘more suitable’. It has to be kept in mind however, that the TTC approach can only be applied when a conventional risk assessment is not possible. Some advantages and limitations of the presented extrapolation

and risk assessment methods are summarized in Table 4.

5. Summary Chemical contaminants and residues in food, air, soil, water and medicinal drugs such as heavy metals, dioxins, pesticide residues etc. may represent a constant but weakly defined risk to human

Table 4 Advantages (þ) and limitations () of the presented point of departures, extrapolation methods and risk assessment approaches according to Davis et al. (2011); COC (2014); SCCS (2012); EFSA (2012). Point of departure/ extrapolation method/risk assessment approach

þ

e

NOAEL/LOAEL

- easy to derive - can be used when a BMD modeling cannot be performed - standard for decades

-

T25 (most applies also for the - quick and easy to calculate TD50 approach) - no elaborate statistics needed - minimum data requirements are one incidence level significantly greater than the controls - allows a relative ranking of different genotoxic carcinogens - applicable even in the case of high tumor incidences in all dose groups, i.e. when no BMD can be derived (but is less meaningful in this case) BMDL - preferred and most recommended method - less dependent on experimental dosages and study design - takes all data points, i.e. the shape of the dose-response curve into account - less conservative due to non-linear extrapolation - better inclusion of variabilities and uncertainties ALARA - easily adapted - easy to communicate MoE - recommended method e.g. for the assessment of genotoxic food constituents - allows a ranking of genotoxic compounds - relative easy to communicate for risk management TTC - easy to apply - needs the knowledge of only two things: chemical structure and exposure level - applicable in case of complete lacking of toxicological data

-

-

not applicable as a POD to genotoxic carcinogens quantized due to dose selection thus may overestimate risk strongly dependent on sample size lacks dose-response information sometimes the lowest dosage group shows effects, i.e. only a LOAEL instead of a NOAEL can be derived no scientific justification for a LOAEL/NOAEL extrapolation tends to generate lower health-based guidance values for studies with higher power to detect adverse effects, thus ‘penalizing’ better studies may overestimate risk due to a more conservative linear interpolation (or extrapolation) many uncertainties not taken into account derivation is more influenced by study design and quality one point linearization (from the lowest tumorigenic dose), i.e. ‘loss’ of information

more time consuming and thus more expensive more complex decision making process not always applicable due to quantity and format of data not applicable when tumor incidences are high in all dosage groups

- no toxicological deviation or justification - values do not (necessarily) reflect a quantitative risk - the cut-off of 10 000 forces the perception to categorize compounds in ‘hazardous’ (MoE < 10 000) and ‘safe’ (MoE > 10 000) which need to be carefully addressed in the risk communication - not permissible when toxicological data are available - generic read-across approach

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

8

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9

health. For an appropriate toxicological risk assessment a variety of elaborated tools are available. In modern toxicology, a crucial prerequisite for a rational risk assessment is a scientific understanding of the compound's mode of action (MoA). For chemicals with a nongenotoxic MoA, the existence of an ineffective threshold for adverse outcomes (with the exception of allergic reactions) for the human population is assumed. In contrast, for chemicals with a genotoxic MoA, in particular if they were carcinogenic in humans or experimental animals, the existence of a threshold and thus of an absolutely safe level cannot be assumed. A need for a risk assessment of genotoxic compounds is rare for (legal) residues, since in most cases genotoxicity is a criterion for exclusion of a legal approval (and thus use) of a chemical. In contrast, contaminants may be genotoxic (carcinogens) and are not always completely unavoidable in consumer products, so that risk assessment approaches for this class of compounds had (and have) to be developed. Here, the Margin of Exposure (MoE) approach describes the distance between the actual exposure in the cohort of interest and the socalled Point of Departure (POD). The latter is a critical benchmark dose level which caused a certain enhanced amount of tumors in an animal model based on quantitative dose-response models. The size of the MoE is used to express the degree of concern towards the public, eventually leading to risk management measures of different stringency. An alternative approach also taking into account the distance between a critical tumorigenic dose in animals and an actual exposure level is the Threshold of Toxicological Concern (TTC) approach for low level compounds, even with critical alerts for genotoxicity/carcinogenicity, but without experimental data sufficient for a conventional risk assessment. Both methods have been applied successfully but are also subject to misconceptions and misuses. Also the concept of the determination of Adverse Outcome Pathways (AOPs, at present under development by OECD) is a promising approach to overcome shortcomings in current risk assessment using certain in vitro test batteries. However, there are still issues which cannot be addressed adequately in risk assessment. Examples are lack of data on the mode of action, inadequate animal data, or the need for a combined risk assessment taking into account exposure to multiple toxicants as well as the risk assessment of only few compounds within complex mixtures or matrices (which may provoke synergistic or antagonistic effects, alter absorption, metabolism etc.), even though there are promising developments especially with regard to novel in silico approaches. Conflict of interest The authors declare that there are no conflicts of interest. Notes The authors declare no competing financial interest. Transparency document Transparency document related to this article can be found online at http://dx.doi.org/10.1016/j.fct.2016.09.012. References Abaci, H.E., Shuler, M.L., 2015. Human-on-a-chip design strategies and principles for physiologically based pharmacokinetics/pharmacodynamics modeling. Integr. Biol. (Camb) 7, 383e391. Albertini, R.J., 2001. HPRT mutations in humans: biomarkers for mechanistic studies. Mutat. Res. 489, 1e16.
pe
e, N., Bahinski, A., Daneshian, M., De Wever, B., Fritsche, E., Goldberg, A., Ale Hansmann, J., Hartung, T., Haycock, J., Hogberg, H., Hoelting, L., Kelm, J.M.,

Kadereit, S., McVey, E., Landsiedel, R., Leist, M., Lübberstedt, M., Noor, F., Pellevoisin, C., Petersohn, D., Pfannenbecker, U., Reisinger, K., Ramirez, T., €fer-Korting, M., Zeilinger, K., Zurich, M.G., 2014. Rothen-Rutishauser, B., Scha State-of-the-art of 3D cultures (organs-on-a-chip) in safety testing and pathophysiology. ALTEX 31, 441e477. http://dx.doi.org/10.14573/altex1406111. Allgaier, C., Franz, S., 2015. Risk assessment on the use of herbal medicinal products containing pyrrolizidine alkaloids. Regul. Toxicol. Pharmacol. 73, 494e500. http://dx.doi.org/10.1016/j.yrtph.2015.09.024. Ames, B.N., Durston, W.E., Yamasaki, E., Lee, F.D., 1973. Carcinogens are mutagens: a simple test system combining liver homogenates for activation and bacteria for detection. Proc. Natl. Acad. Sci. U. S. A. 70, 2281e2285. An, F., Qu, Y., Liu, X., Zhong, R., Luo, Y., 2015. Organ-on-a-chip: new platform for biological analysis. Anal. Chem. Insights 10, 39e45.
Júnior, P.L., Nozima, B.H., Ito, E.T., de Araldi, R.P., de Melo, T.C., Mendes, T.B., de Sa Carvalho, R.F., de Souza, E.B., de Cassia Stocco, R., 2015. Using the comet and micronucleus assays for genotoxicity studies: a review. Biomed. Pharmacother. 72, 74e82. http://dx.doi.org/10.1016/j.biopha.2015.04.004. Balbo, S., Turesky, R.J., Villalta, P.W., 2014. DNA adductomics. Chem. Res. Toxicol. 27, 356e366. http://dx.doi.org/10.1021/tx4004352. Bernard, A., Broeckaert, F., De Poorter, G., De Cock, A., Hermans, C., Saegerman, C., Houins, G., 2002. The Belgian PCB/dioxin incident: analysis of the food chain contamination and health risk evaluation. Environ. Res. 88, 1e18. Blaauboer, B.J., Boobis, A.R., Bradford, B., Cockburn, A., Constable, A., Daneshian, M., Edwards, G., Garthoff, J.A., Jeffery, B., Krul, C., Schuermans, J., 2016. Considering new methodologies in strategies for safety assessment of foods and food ingredients. Food Chem. Toxicol. 91, 19e35. Bucher, J.R., 2002. The National Toxicology Program rodent bioassay: designs, interpretations, and scientific contributions. Ann. N. Y. Acad. Sci. 982, 198e207. Bundesamt für Risikobewertung (BfR; German Federal Institute for Risk Assessment), 2005. Risk Assessment of Genotoxic and Carcinogenic Substances to Be Harmonised in the EU. BfR Expert Opinion No. 029/2005 of 18 May 2005. Available at: http://www.bfr.bund.de/cm/343/harmonised_approach_for_the_ risk_assessment_of_compounds_which_are_both_genotoxic_and_carcinogenic. pdf (Accessed 24.05.16). Burden, N., Mahony, C., Müller, B.P., Terry, C., Westmoreland, C., Kimber, I., 2015. Aligning the 3Rs with new paradigms in the safety assessment of chemicals. Toxicology 330, 62e66. Burm, E., Ha, M., Kwon, H.J., 2015. Association between blood cadmium level and bone mineral density reduction modified by renal function in young and middle-aged men. J. Trace Elem. Med. Biol. 32, 60e65. http://dx.doi.org/ 10.1016/j.jtemb.2015.06.002. Cheeseman, M.A., Machuga, E.J., Bailey, A.B., 1999. A tiered approach to threshold of regulation. Food Chem. Toxicol. 37, 387e412. Committee on Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (COC), 2014. Defining a Point of Departure and Potency Estimates in Carcinogenic Dose Response. COC/G 05 e Version 1.0. Available at: https:// www.gov.uk/government/uploads/system/uploads/attachment_data/file/ 359324/Defining_a_point_of_departure_and_potency_estimates_in_ carcinogenic_dose_response.pdf (Accessed 25.05.16). Codex Alimentarius Commission, 2015. Procedural Manual, 24th edition. FAO/WHO, Rome. ISSN 1020e8070. Available at: http://www.fao.org/fao-whocodexalimentarius/procedures-strategies/procedural-manual/en/. (Accessed 27.06.16). Corvi, R., Madia, F., 2016. In vitro genotoxicity testing - Can the performance be enhanced? Food Chem.Toxicol. http://dx.doi.org/10.1016/j.fct.2016.08.024. Cramer, G.M., Ford, R.A., Hall, R.L., 1978. Estimation of toxic hazardea decision tree approach. Food Cosmet. Toxicol. 16, 255e276. Davis, J.A., Gift, J.S., Zhao, Q.J., 2011. Introduction to benchmark dose methods and U.S. EPA's benchmark dose software (BMDS) version 2.1.1. Toxicol. Appl. Pharmacol. 254, 181e191. http://dx.doi.org/10.1016/j.taap.2010.10.016. Diggle, G.E., 2001. The toxic oil syndrome: 20 years on. Int. J. Clin. Pract. 55, 371e375. European Commission (EC), 2012. SCENIHR (Scientific Committee on Emerging and Newly Identified Health Risks), SCHER (Scientific Committee on Health and Environmental Risks), SCCS (Scientific Committee on Consumer Safety), Addressing the New Challenges for Risk Assessment, 8 October 2012. Available at: http://www.ec.europa.eu/health/scientific_committees/emerging/docs/ scenihr_o_037.pdf (Accessed 20.05.16). European Food Safety Authority (EFSA), 2011. Scientific Opinion on Pyrrolizidine alkaloids in food and feed. EFSA J. 9 (11), 2406. http://dx.doi.org/10.2903/ j.efsa.2011.2406. Available at: https://www.efsa.europa.eu/sites/default/files/ scientific_output/files/main_documents/2406.pdf (Accessed 24.05.16). European Food Safety Authority (EFSA), 2011b. Scientific Opinion on the risks for animal and public health related to the presence of Alternaria toxins in feed and food. EFSA J. 9 (10), 2407. http://dx.doi.org/10.2903/j.efsa.2011.2407. Available at: http://www.efsa.europa.eu/sites/default/files/scientific_output/files/main_ documents/2407.pdf (Accessed 25.05.16). European Food Safety Authority (EFSA), 2012. Guidance on the use of probabilistic methodology for modelling dietary exposure to pesticide residues. EFSA J. 10 (10), 2839. http://dx.doi.org/10.2903/j.efsa.2012.2839. Available at: http:// www.efsa.europa.eu/sites/default/files/scientific_output/files/main_ documents/2839.pdf (Accessed 20.05.16). European Food Safety Authority (EFSA), 2012b. Scientific Opinion on Exploring options for providing advice about possible human health risks based on the concept of Threshold of Toxicological Concern (TTC). EFSA J. 10 (3), 2578. http://

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012

A. Cartus, D. Schrenk / Food and Chemical Toxicology xxx (2016) 1e9 dx.doi.org/10.2903/j.efsa.2012.2578. Available at: https://www.efsa.europa.eu/ de/efsajournal/pub/2578 (Accessed 24.05.16). European Food Safety Authority (EFSA), 2012c. Statement on the applicability of the Margin of Exposure approach for the safety assessment of impurities1 which are both genotoxic and carcinogenic in substances added to food/feed. EFSA J. 10 (7), 2750. http://dx.doi.org/10.2903/j.efsa.2012.2750. Available at: http:// www.efsa.europa.eu/sites/default/files/scientific_output/files/main_ documents/2750.pdf (Accessed 25.05.16). European Food Safety Authority (EFSA), 2015. Scientific Opinion on acrylamide in food. EFSA J. 13 (6), 4104. http://dx.doi.org/10.2903/j.efsa.2015.4104. Available at: http://www.efsa.europa.eu/sites/default/files/scientific_output/files/main_ documents/4104.pdf (Accessed 24.05.16). European Food Safety Authority (EFSA), 2016. The EFSA Comprehensive European Food Consumption Database. Available at: https://www.efsa.europa.eu/en/ food-consumption/comprehensive-database. € rsam, B., Thomas, A.D., Reißig, S., Fahrer, J., Frisch, J., Nagel, G., Kraus, A., Do Waisman, A., Kaina, B., 2015. DNA repair by MGMT, but not AAG, causes a threshold in alkylation-induced colorectal carcinogenesis. Carcinogenesis 36, 1235e1244. http://dx.doi.org/10.1093/carcin/bgv114. Farmer, P.B., Singh, R., 2008. Use of DNA adducts to identify human health risk from exposure to hazardous environmental pollutants: the increasing role of mass spectrometry in assessing biologically effective doses of genotoxic carcinogens. Mutat. Res. 659, 68e76. http://dx.doi.org/10.1016/j.mrrev.2008.03.006. Fowler, P., Smith, R., Smith, K., Young, J., Jeffrey, L., Kirkland, D., Pfuhler, S., Carmichael, P., 2012. Reduction of misleading (“false”) positive results in mammalian cell genotoxicity assays. II. Importance of accurate toxicity measurement. Mutat. Res. 747, 104e117. http://dx.doi.org/10.1016/ j.mrgentox.2012.04.013. Fukushima, S., Gi, M., Kakehashi, A., Wanibuchi, H., Matsumoto, M., 2016. Qualitative and quantitative approaches in the dose-response assessment of genotoxic carcinogens. Mutagenesis 31, 341e346. http://dx.doi.org/10.1093/mutage/ gev049. Hayashi, M., Dearfield, K., Kasper, P., Lovell, D., Martus, H.J., Thybaud, V., 2011. Compilation and use of genetic toxicity historical control data. Mutat. Res. 723, 87e90. Keith, G., Dirheimer, G., 1995. Postlabeling: a sensitive method for studying DNA adducts and their role in carcinogenesis. Curr. Opin. Biotechnol. 6, 3e11. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J.G., Würtzen, G., 2004. Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65e83. Kroes, R., Renwick, A.G., Feron, V., Galli, C.L., Gibney, M., Greim, H., Guy, R.H., Lhuguenot, J.C., van de Sandt, J.J., 2007. Application of the threshold of toxicological concern (TTC) to the safety evaluation of cosmetic ingredients. Food Chem. Toxicol. 45, 2533e2562. Lagarde, F., Beausoleil, C., Belcher, S.M., Belzunces, L.P., Emond, C., Guerbet, M., Rousselle, C., 2015. Non-monotonic dose-response relationships and endocrine disruptors: a qualitative method of assessment. Environ. Health 14, 13. http:// dx.doi.org/10.1186/1476-069X-14-13.
, T., Riviere, J.E., 2016. Mathematical modeling Lin, Z., Gehring, R., Mochel, J.P., Lave and simulation in animal health - Part II: principles, methods, applications, and value of physiologically based pharmacokinetic modeling in veterinary medicine and food safety assessment. J. Vet. Pharmacol. Ther. 39, 421e438. http:// dx.doi.org/10.1111/jvp.12311. Madle, S., Dean, S.W., Andrae, U., Brambilla, G., Burlinson, B., Doolittle, D.J., Furihata, C., Hertner, T., McQueen, C.A., Mori, H., 1994. Recommendations for the performance of UDS tests in vitro and in vivo. Mutat. Res. 312, 263e285. Maronpot, R.R., Nyska, A., Foreman, J.E., Ramot, Y., 2016. The legacy of the F344 rat as a cancer bioassay model (a retrospective summary of three common F344 rat neoplasms). Crit. Rev. Toxicol. 46, 641e675. http://dx.doi.org/10.1080/ 10408444.2016.1174669. Mortelmans, K., Zeiger, E., 2000. The Ames Salmonella/microsome mutagenicity assay. Mutat. Res. 455, 29e60. Nath, S., Devi, G.R., 2016. Three-dimensional culture systems in cancer research: focus on tumor spheroid model. Pharmacol. Ther. 163, 94e108. http:// dx.doi.org/10.1016/j.pharmthera.2016.03.013 pii: S0163e7258(16)30021-3. Nesslany, F., 2016. The current limitations of in vitro genotoxicity testing and their relevance to the in vivo situation. Food Chem. Toxicol. http://dx.doi.org/10.1016/ j.fct.2016.08.035.

9

Peto, A.E., Pike, M.C., Bernstein, L., Gold, L.S., Ames, B.N., 1984. The TD50: a proposed general convention for the numerical description of the carcinogenic potency of chemicals in chronic-exposure animal experiments. Environ. Health Perspect. 58, 1e8. Rani, A., Kumar, A., Lal, A., Pant, M., 2014. Cellular mechanisms of cadmium-induced toxicity: a review. Int. J. Environ. Health Res. 24, 378e399. http://dx.doi.org/ 10.1080/09603123.2013.835032. Re, T.A., Mooney, D., Antignac, E., Dufour, E., Bark, I., Srinivasan, V., Nohynek, G., 2009. Application of the threshold of toxicological concern approach for the safety evaluation of calendula flower (Calendula officinalis) petals and extracts used in cosmetic and personal care products. Food Chem. Toxicol. 47, 1246e1254. http://dx.doi.org/10.1016/j.fct.2009.02.016. Reifferscheid, G., Maes, H.M., Allner, B., Badurova, J., Belkin, S., Bluhm, K., Brauer, F., Bressling, J., Domeneghetti, S., Elad, T., Flückiger-Isler, S., Grummt, H.J., Gürtler, R., Hecht, A., Heringa, M.B., Hollert, H., Huber, S., Kramer, M., Magdeburg, A., Ratte, H.T., Sauerborn-Klobucar, R., Sokolowski, A., Soldan, P., Smital, T., Stalter, D., Venier, P., Ziemann, C., Zipperle, J., Buchinger, S., 2012. International round-robin study on the Ames fluctuation test. Environ. Mol. Mutagen. 53, 185e197. http://dx.doi.org/10.1002/em.21677. Rhomberg, L.R., Baetcke, K., Blancato, J., Bus, J., Cohen, S., Conolly, R., Dixit, R., Doe, J., Ekelman, K., Fenner-Crisp, P., Harvey, P., Hattis, D., Jacobs, A., Jacobson-Kram, D., Lewandowski, T., Liteplo, R., Pelkonen, O., Rice, J., Somers, D., Turturro, A., West, W., Olin, S., 2007. Issues in the design and interpretation of chronic toxicity and carcinogenicity studies in rodents: approaches to dose selection. Crit. Rev. Toxicol. 37, 729e837. Safford, R.J., 2008. The Dermal Sensitisation Threshold- a TTC approach for allergic contact dermatitis. Regul. Toxicol. Pharmacol. 51, 195e200. Sand, S., Portier, C.J., Krewski, D., 2011. A signal-to-noise crossover dose as the point of departure for health risk assessment. Environ. Health Perspect. 119, 1766e1774. http://dx.doi.org/10.1289/ehp.1003327. Sanner, T., Dybing, E., Willems, M.I., Kroese, E.D., 2001. A simple method for quantitative risk assessment of non-threshold carcinogens based on the dose descriptor T25. Pharmacol. Toxicol 88, 331e341. Santella, R.M., 1999. Immunological methods for detection of carcinogen-DNA damage in humans. Cancer Epidemiol. Biomarkers Prev. 8, 733e739. Scientific Committee on Consumer Safety (SCCS)/Scientific Committee on Health and Environmental Risks (SCHER)/Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR), 2012. Opinion on: Use of the Threshold of Toxicological Concern (TTC) Approach for Human Safety Assessment of Chemical Substances with Focus on Cosmetics and Consumer Products. SCCP/1171/08. Available at: http://ec.europa.eu/health/scientific_committees/ consumer_safety/docs/sccs_o_092.pdf (Accessed 25.05.16). Smith, B., Cadby, P., Leblanc, J.C., Setzer, R.W., 2010. Application of the margin of exposure (MoE) approach to substances in food that are genotoxic and carcinogenic: example: methyleugenol, CASRN: 93-15-2. Food Chem. Toxicol. 48 (Suppl. 1), S89eS97. http://dx.doi.org/10.1016/j.fct.2009.10.036. Sung, J.H., Srinivasan, B., Esch, M.B., McLamb, W.T., Bernabini, C., Shuler, M.L., Hickman, J.J., 2014. Using physiologically-based pharmacokinetic-guided “body-on-a-chip” systems to predict mammalian response to drug and chemical exposure. Exp. Biol. Med. (Maywood) 239, 1225e1239. Szekely, G., Amores de Sousa, M.C., Gil, M., Castelo Ferreira, F., Heggie, W., 2015. Genotoxic impurities in pharmaceutical manufacturing: sources, regulations, and mitigation. Chem. Rev. 115, 8182e8229. Thomas, A.D., Fahrer, J., Johnson, G.E., Kaina, B., 2015. Theoretical considerations for thresholds in chemical carcinogenesis. Mutat. Res. Rev. Mutat. Res. 765, 56e67. http://dx.doi.org/10.1016/j.mrrev.2015.05.001. Vinken, M., 2013. The adverse outcome pathway concept: a pragmatic tool in toxicology. Toxicology 312, 158e165. Wei, M., Kakehashi, A., Yamano, S., Tamano, S., Shirai, T., Wanibuchi, H., Fukushima, S., 2012. Lack of hepatocarcinogenicity of combinations of low doses of 2-amino-3, 8-dimethylimidazo[4,5- f]quinoxaline and diethylnitrosamine in rats: indication for the existence of a threshold for genotoxic carcinogens. J. Toxicol. Pathol. 25, 209e214. http://dx.doi.org/10.1293/tox.25.209. Williams, F.M., Rothe, H., Barrett, G., Chiodini, A., Whyte, J., Cronin, M.T., MonteiroRiviere, N.A., Plautz, J., Roper, C., Westerhout, J., Yang, C., Guy, R.H., 2016. Assessing the safety of cosmetic chemicals: consideration of a flux decision tree to predict dermally delivered systemic dose for comparison with oral TTC (Threshold of Toxicological Concern). Regul. Toxicol. Pharmacol. 76, 174e186. http://dx.doi.org/10.1016/j.yrtph.2016.01.005.

Please cite this article in press as: Cartus, A., Schrenk, D., Current methods in risk assessment of genotoxic chemicals, Food and Chemical Toxicology (2016), http://dx.doi.org/10.1016/j.fct.2016.09.012