Scholarly article on topic 'Approaches for grouping of pesticides into cumulative assessment groups for risk assessment of pesticide residues in food'

Approaches for grouping of pesticides into cumulative assessment groups for risk assessment of pesticide residues in food Academic research paper on "Veterinary science"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"Cumulative risk assessment" / Pesticides / Organophosphates / Thyroid / "Margin of exposure" / "Human relevance" / "Nervous system" / "Cumulative assessment groups (CAGs)"}

Abstract of research paper on Veterinary science, author of scientific article — Thomas Colnot, Wolfgang Dekant

Abstract The European Food Safety Authority (EFSA) is developing approaches to cumulative risk assessment of pesticides by assigning individual pesticides to cumulative assessment groups (CAGs). For assignment to CAGs, EFSA recommended to rely on adverse effects on the specific target system. Contractors to EFSA have proposed to allocate individual pesticides into CAGs relying on NOAELs for effects on target organs. This manuscript evaluates the assignments by applying EFSAs criteria to the CAGs “Toxicity to the nervous system” and “Toxicity to the thyroid hormone system (gland or hormones)”. Assignment to the CAG “Toxicity to the nervous system” based, for example, on neurochemical effects like choline esterase inhibition is well supported, whereas assignment to the CAG “Toxicity to the thyroid hormone system (gland or hormones)” has been based in the examined case studies on non-reproducible effects seen in single studies or on observations that are not adverse. Therefore, a more detailed effects evaluation is required to assign a pesticide to a CAG for a target organ where many confounders regarding effects are present. Relative potency factors in cumulative risk assessment should be based on benchmark doses from studies in one species with identical study design and human relevance of effects on specific target organs should be analyzed to define minimal margins of exposure.

Academic research paper on topic "Approaches for grouping of pesticides into cumulative assessment groups for risk assessment of pesticide residues in food"

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology

journal homepage: www.elsevier.com/locate/yrtph

Approaches for grouping of pesticides into cumulative assessment groups for risk assessment of pesticide residues in food

CrossMark

Thomas Colnot a, Wolfgang Dekant

a CiS Toxicology, Osorno, Chile

b Department of Toxicology, University of Wurzburg, Versbacher Strasse 9, 97078, Wiirzburg, Germany

ARTICLE INFO

Article history:

Received 14 October 2016

Received in revised form

6 December 2016

Accepted 7 December 2016

Available online 10 December 2016

Keywords:

Cumulative risk assessment Pesticides Organophosphates Thyroid

Margin of exposure Human relevance Nervous system

Cumulative assessment groups (CAGs)

ABSTRACT

The European Food Safety Authority (EFSA) is developing approaches to cumulative risk assessment of pesticides by assigning individual pesticides to cumulative assessment groups (CAGs). For assignment to CAGs, EFSA recommended to rely on adverse effects on the specific target system. Contractors to EFSA have proposed to allocate individual pesticides into CAGs relying on NOAELs for effects on target organs. This manuscript evaluates the assignments by applying EFSAs criteria to the CAGs "Toxicity to the nervous system" and "Toxicity to the thyroid hormone system (gland or hormones)". Assignment to the CAG "Toxicity to the nervous system" based, for example, on neurochemical effects like choline esterase inhibition is well supported, whereas assignment to the CAG "Toxicity to the thyroid hormone system (gland or hormones)" has been based in the examined case studies on non-reproducible effects seen in single studies or on observations that are not adverse. Therefore, a more detailed effects evaluation is required to assign a pesticide to a CAG for a target organ where many confounders regarding effects are present. Relative potency factors in cumulative risk assessment should be based on benchmark doses from studies in one species with identical study design and human relevance of effects on specific target organs should be analyzed to define minimal margins of exposure.

© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

EFSA has recommended the concept of dose addition for risk characterization of mixtures of pesticides even for pesticides with a dissimilar mode of action if they produce a common and adverse outcome on a target organ/system (EFSA-PPR-Panel, 2013b, 2014). This is a conservative and pragmatic default approach since mode of action data regarding induction of adverse effects are often limited (EC-SCHER, 2012). Therefore, individual pesticides inducing a common adverse outcome in the same target organ/system should be grouped into „common assessment groups" (CAGs) for cumulative human health risk characterization. The tiered grouping approach consists of several levels, level 1 is a common target organ and level 2 specifically describes the phenomenolog-ical effect (EFSA-PPR-Panel, 2013b, 2014). At levels 3 and 4, additional mode and/or mechanism of action information could be integrated. However, such information is not available for the majority of pesticides. Thus, grouping into CAGs is currently based on

* Corresponding author. E-mail address: dekant@toxi.uni-wuerzburg.de (W. Dekant).

phenomenological effects at level 2.

Cumulative risk assessment requires both identification of relevant adverse effects on the target organ/system and a comparative assessment of potency of the individual compounds in a CAG. Identification of relevant toxic effects requires a thorough hazard assessment considering information from all toxicity studies available and its consistency. Relative potency may be expressed by No-Observed-Adverse-Effect-Levels" (NOAELs) or benchmark doses (BMDs) for the common and adverse outcome. The relative potency of individual members in a CAG, in combination with exposure characteristics, may have significant consequences for tolerable maximum residue levels for individual pesticides in food. Therefore, assignment of pesticides to CAGs and assessment of relative potency have to be based on the best available science using the most appropriate methodology (Kienzler et al., 2016).

Regarding inclusion of a pesticide into a target organ/system specific CAG, EFSA has developed a methodology to identify effects of an individual pesticide on target organs/systems. The methodology proposes a stepwise approach with a major focus on identification and characterization of the specific effect. EFSA apparently

http://dx.doi.org/10.1016/j.yrtph.2016.12.004

0273-2300/© 2016 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

relies on systemic "adverse effects" as defined by IPCS (WHO/IPCS, 2004) since "local effects", "non-adverse effects", and "non-specific effects" should be excluded. In addition, only "unambiguous effects" (EFSA-PPR-Panel, 2013a) interpreted here as "clear adverse effects of sufficient magnitude and biological relevance as a consequence of administration of a specific chemical" should be considered. Effects without human relevance should also not be considered. This is a reasonable approach following well-established procedures (EFSA-PPR-Panel, 2014; Dekant and Bridges, 2016).

Contractors to EFSA have already proposed to include a number of active pesticides into specific CAGs (level 2) with a focus on the nervous system and the thyroid (supplementary material, Table 1 s). The nervous system CAG was further subdivided into CAGs for acute effects on motor, sensory, and autonomic divisions of the nervous system, and into neurochemical endpoints. Pesticides with effects on the thyroid were allocated to CAGs for effects either on C-cells and the calcitonin system or to a group affecting thyroid hormones (T3/T4) and the thyroid follicular cells. The CAG-group "thyroid follicular cell hypertrophy/hyperplasia and/or increased relative thyroid weight" has the largest number of pesticides included (EFSA-PPR-Panel, 2013b).

Effects to be used as a basis for assignment to a CAG and organ-specific NOAELs and LOAELs were derived by contractors to EFSA. In this context, the "selection of NOAELs and LOAELs was performed, as requested by EFSA, without any interpretation of whether an effect is to be considered as adverse or not adverse" implying that any effect reported was used. Apparently, there was no consideration of study quality, dose-response, or consistency of effects over studies. This procedure is inconsistent with the practice of toxicological risk characterization and with the basic approach outlined by EFSA (EFSA-PPR-Panel, 2014). Assessment of adversity and unambiguity of an effect requires a detailed evaluation of the database and cannot be limited to giving NOAELs/LOAELs.

Apparently, this very wide definition served as an initial screening only and requires further consideration along the lines outlined in the guidance documents. Therefore, this manuscript evaluates the results of the grouping of pesticides into CAGs with focus on the level 2 CAGs "thyroid follicular cell/T3/4 system" and "neurochemical effects" regarding support for inclusion of a pesticide into the respective CAG using a selected set of pesticides (cyhalofop-butyl, dithianon, ametoctradin, amidosulfuron) (EFSA-PPR-Panel, 2014; RIVM-ICPS-ANSES, 2016). Datasets for these pesticides were evaluated by applying the criteria of weight-of-evidence and adversity based on the European Peer Review conclusions, the Draft Assessment Reports (DARs) and original study reports. In addition, the manuscript proposes approaches to derive robust potency factors based on adverse effects to serve as a basis for the cumulative assessment. Our approach is in line with EFSAs SC recommendation for risk characterization that focus on "unambiguous" and "adverse effects" (EFSA-PPR-Panel, 2013a).

2. Methodology to select appropriate studies as a basis for assignment of pesticides to the level 2 CAGs "neurochemical group" and "thyroid follicular cell/T3/T4 system"

2.1. General criteria for study evaluation

A solid assessment of the hazardous properties of a chemical requires both a quality assessment of the respective study to judge the reliability of the study outcome and an evaluation of the toxic effects reported to judge their adversity and consistency regarding different levels of the biological response from biomarkers to functional and structural changes. In many cases, quality assessment of a toxicity study is based on an application of the Klimisch

scale or other rather superficial approaches (Klimisch et al., 1997; Lutter et al., 2015). The major limitation of the Klimisch scale is that "reliability without restrictions" is only allocated to studies performed following OECD study guidelines with full study reports containing raw data. Other studies including those published in the peer-reviewed scientific literature are termed "reliable with restrictions". However, the restrictions regarding reliability are not explicitly defined and a systematic approach to determine reliability is not included. Therefore, the criteria of the Klimisch scale focus more on the quality of reporting and not the quality of the actual work. For this reason, a more detailed quality assessment is required (Dekant and Bridges, 2016). Quality assessment needs to include the extent of characterization of the test chemical, its stability in the application medium, and potential presence of contaminants. Evaluation of the study design needs to consider the species selected, the number of animals/dose group and the number and spacing of dose levels, the appropriateness of the doserange tested, the inclusion of appropriate controls, and the relevance of the route and timing of administration. Other important issues are the reliability of the applied methodology for analytical measurements (such as thyroid hormones or cholinesterase activity) and potential issues with sample preparation for histopathol-ogy. Finally, the appropriateness of procedures for statistical evaluation of incidence data needs to be evaluated and, in case of equivocal effects, compared to historical control data for the selected endpoints.

Effects assessment also requires a detailed analysis of the reported changes. As EFSA apparently proposes to use "adverse effects" as a basis to include specific chemicals into CAGs, adversity of all changes reported needs to be determined. A critical issue in the determination of adversity consists in the question if changes are really adverse and biologically relevant, represent adaptive or physiological changes, or are random events not related to treatment. Adverse effects are defined by IPCS as those "that result in an impairment of functional capacity, an impairment of the capacity to compensate for additional stress, or an increase in susceptibility to other influences" (WHO/IPCS, 2004), while non-adverse or adaptive effects are those "biochemical, morphological, or physiological changes that do not affect the general well-being, growth, development or life span" (Lewis et al., 2002).

To qualify as adverse, effects reported in a study usually are required to exhibit a dose-response and be consistent with the other changes regarding hypothesis of disease development, e.g. an adverse outcome pathway (Carmichael et al., 2011; Simon et al., 2014; Sturla et al., 2014). In this analysis, a change in a parameter is considered adverse if it shows a dose response with higher incidences/intensity at higher doses. Moreover, when biomarkers such as changes in enzyme activity or hormone levels are used to derive a point-of-departure (POD) for risk characterization, the change in the biomarker is considered as biologically significant only when adverse effects were observed at higher doses or after longer exposures. Assessment of the relevance of changes in bio-markers and adverse effects on a target organ also requires to consider the possibility that reported changes are secondary to other effects induced by the treatment. For example, when there are clear indications of liver changes such as increased liver weight due to induction of biotransformation enzymes, deriving a potency factor regarding thyroid changes at higher doses is inappropriate since such changes are a consequence of the modified disposition of thyroid hormones that only occurs after administration of doses that result in significant induction of xenobiotic metabolizing enzymes. In case of conflicting information, a detailed weight-of-evidence analysis is required to decide if a biomarker changes should be used as a basis of a POD (Lamb et al., 2015; Lutter et al., 2015).

2.2. Specific criteria for neurochemical effects and thyroid hormones/thyroid follicular cells

2.2.1. Neurochemical effects group

With regard to the nervous system, four acute (functional effects on motor division, functional effects on sensory division, functional effects on autonomic division, neurochemical endpoints) and five chronic groups (functional effects on motor division, functional effects on sensory division, functional effects on autonomic division, neurochemical endpoints, neuropathological endpoints) were proposed (EFSA-PPR-Panel, 2014). In an external report (RIVM-ICPS-ANSES, 2016), two further "acute groups" (neuropathological endpoints, developmental neurotoxicity) and one further "chronic group" (developmental neurotoxicity) were added. Some of the pesticides proposed for grouping into specific CAGs are inadequately supported and two examples for inappropriate grouping are discussed below.

2.2.1.1. Acute effects group. Fluquinconazole is proposed to be grouped in the acute subgroup "motor division" based on tremors observed at 1.79 mg/kg bw (EFSA, 2013) and a NOAELof 0.45 mg/kg bw. (Level 1), A detailed evaluation of the database of fluquinco-nazole shows that tremor in females at a dose of 1.79 mg/kg bw is observed only on day 24. Thus, tremors are not an acute, but a chronic endpoint and assignment of fluquinconazole to the acute neurotoxicity subgroup "motor division" is inappropriate. Fluo-pyram was also proposed to be included in the subgroup "motor division" of the target organ "nervous system" based on reduced motor activity seen in an acute neurotoxicity study in rats at a dose level of 100 mg/kg bw. The NOAEL in this study was 50 mg/kg bw. In the European Peer Review, this effect was not considered as an indication for a specific neurotoxicity and EFSA concluded, that "Fluopyram did not show any specific potential for neurotoxicity" (EFSA, 2013). Therefore, the grouping for Fluopyram into the "motor division" CAG also requires revision.

2.2.1.2. Chronic effects group. The chronic neurochemical effects group has been selected as an example for detailed analysis since half of the active pesticides allocated to this group have a proposed target organ-specific NOAEL lower than the NOAEL used as a basis

for the ADI (Table 1). This seems inappropiate as, consequently, the ADI is no longer be supported. Organophosphates containing a leaving group and carbamates are widely used pesticides and exhibit a specific mode of action, inhibition of acetylcholinesterase (AChE) by phosphorylation or carbamylation. This results in an accumulation of acetylcholine in the nervous system with consequent overstimulation at muscarinic and nicotinic cholinergic synapses. Overstimulation at muscarinic synapses results in hy-persalivation, excess lacrimation, miosis, and bronchoconstriction and overstimulation at nicotinic synapses results in muscle cramps, fasciculation, weakness, and paralysis. Central nervous system (CNS) effects include anxiety, convulsions and respiratory depression (Ballantyne et al., 2009; Klaassen, 2013). Severity of symptoms and time course depend on the chemical structure, the dose, the route, the frequency and duration of exposure, and the time of observation relative to the time of peak toxic effect. AChE-activity can be readily determined in blood with simple methods (Chambers et al., 2010) and blood AChE-activity serves as a biomarker of AChE-activity in the CNS that is directly related to the adverse effects. Due to the clearly defined effect that represents the intended effect against pests, the specific biomarker AChE-activity, and the structural requirements for binding to AChE consistent with the intended mode-of-action (Ballantyne et al., 2009), grouping of different pesticides into the CAG "neurochemical group" is comparatively simple. EFSA has grouped all pesticides into this group based on AChE-inhibition in blood or brain, which is straightforward and reasonable. However, a variety of studies using different species and study designs were used to derive NOAELs (see Table 1). In addition. endpoints from shorter-term studies were used in many cases instead of the more appropriate longer-term studies. In two cases, an overall weight of evidence evaluation using the results of all available studies had been used to derive the correct NOAELs instead of using isolated endpoints (Phosmet and Dimethoate).

2.2.2. Thyroid hormones/thyroid follicular cells group

EFSA has identified a large number of pesticides as having effects on the thyroid and thus has allocated a large number of pesticides to the level 2 CAG "follicular cells/the T3/T4 system". Again, allocation was performed based on results observed in studies with

Table 1

Members of the chronic neurochemical endpoint group and comparison of NOAELs used in CAG grouping regarding "neurochemical effects" by EFSA (NOAELcag) with those used for deriving acceptable daily intake (NOAEL№I) for the same pesticide based on selected examples. Substances where there was no difference between the NOAELadi and the NOAELCAG are not listed.

Substance ADIa NOAELadi NOAEL for neurochemical Possible rationale for difference; in case NOAEL for target organ is lower, than NOAELadi

[mg/kg bw] endpointb [mg/kg bw]

Pirimiphos- 0.004 0.25 0.25 mg/kg bw is from a 28-day human volunteer study; 0.4 is from the 2-year rat study

methyl 0.4

Phosmet 0.01 <1 1 mg/kg bw is the LOAEL of the mouse carcinogenicity study; the 1 mg/kg bw is considered to be the

1 overall NOEL based on rat and dog studies

Ethoprofos 0.0004 0.025 0.025 mg/kg bw is from a 90-day neurotoxicity based on brain AChE, while 0.04 mg/kg bw is based on a 2-

0.04 year rat study

Chlorpyrifos 0.001 0.03 The NOAEL of 0.03 mg/kg bw is from a 90-day rat and dog study, while the NOAEL of 0.1 mg/kg bw is based

0.1 on 2-year rat and dog studies red blood cell ChE-inhibtion

Dimethoate 0.001 0.04 The NOAEL of 0.04 mg/kg bw derived from a 2-year rat study. The LOAEL of that study was 0.2 mg/kg bw.

0.1 The NOAELs from the studies were 0.1 mg/kg bw, which was considered to be valid for the 2-year chronic

Fenamiphos 0.0008 0.042 The NOAEL of 0.042 mg/kg bw is based on a 90-day dog study, while the NOAEL of 0.08 mg/kg bw is based

0.08 on a 1-year dog study

Methiocarb 0.013 <0.05 The NOAEL of 1.3 mg/kg bw is from a 90-day dog study, while the NOAEL of <0.05 is from a 29-day

1.3 neurotoxicity study

Chlorpyrifos- 0.01 0.1 The NOAEL of 0.1 mg/kg bw is derived from a 28-day human subjects and a 90-day dog study. The 1 mg/kg

methyl 1 bw was derived based on a 1-year rat study

1) Endpoint for ADI setting was revised by EFSA in 2014. a Taken from EC pesticides data base (http://ec.europa.eu/food/plant/pesticides/eu-pesticides-database/public. b Taken from EFSA Scientific Opinion, 2013.

widely different study designs and in different species with apparently little consideration of database consistency and dose-response. Moreover, the identified organ-specific NOAELs/LOAELs for the pesticides to be included into this CAG were also often lower than those used for deriving ADIs (Table 2).

The thyroid is not a common target organ for pesticide action and assessment of thyroid effects regarding their adversity and thus their inclusion into the "follicular cell/the T3/T4 system" CAG requires specific considerations of thyroid function and effects (EFSA-PPR-Panel, 2013b, 2014).

Production and function of thyroid hormones are similar in humans and rodents. Production of3,5,3'-triiodothyronine (T3) and tetraiodothyronine (T4, thyroxine) are regulated by a feedback mechanism controlled by the hypothalamic—pituitary—thyroid (HPT) axis. Low levels of circulating T3 and T4 trigger the hypothalamus to release thyrotropin-releasing hormone (TRH). In turn, TRH stimulates the pituitary to produce thyroid-stimulating hormone (TSH) that stimulates the thyroid to increase the production of thyroid hormone until levels in blood return to normal. Finally, the increase in thyroid hormones exerts a negative feedback control over the hypothalamus as well as anterior pituitary, thus controlling the release of both TRH and TSH (Hurley, 1998; Wu and Farrelly, 2006; Dietrich et al., 2012). More than 99% of the circulating thyroid hormones are bound to plasma proteins, mainly to thyroxine-binding globulin (TGB) in humans and to transthyretin and albumin in rodents. T4 is activated in the liver by deiodination to the hormonally active T3 and excess thyroid hormones are deactivated in the liver by glucuronide formation (Cunha and van Ravenzwaay, 2005).

2.2.2.1. Relevance of changes in thyroid hormone levels regarding adverse effects. Endpoints relevant to this CAG available from repeated-dose toxicity studies may include thyroid gland absolute and relative weight, thyroid gland histopathology, and blood levels of thyroid hormones (T4, T3, and TSH). Weight changes in the thyroid and thyroid histopathology are clearly adverse effects that require further consideration (Hoyer and Flaws, 2013). However, in the absence of functional changes in the thyroid, the question arises if differences in thyroid hormone levels between controls and treatment groups represent an adverse effect. When evaluating the potential adversity of thyroid hormone changes in rats, several caveats should be considered (Lewis et al., 2002):

• The determination of thyroid hormone levels is inherently imprecise and is influenced by many factors. Special care needs to be taken in the interpretation of thyroid hormone results determined in older studies, when such measurements were less frequently performed and analytical methods were not properly validated for a specific animal model. Also, the absence of historical control data early in routinely conducted thyroid hormone measurements makes interpretation difficult. Moreover, thyroid hormone and TSH levels may be influenced by stress (Dohler et al., 1977; Gartner et al., 1980), time of day, estrous cycle, and age of the animals (DeVito et al., 1999).

• A difference in thyroid hormone levels between controls and treated animals is unlikely a treatment-related effect in the absence of dose-response.

• A difference in thyroid hormone levels is unlikely related to treatment if it is due to findings in one or a few animals.

• A difference in thyroid hormone levels is less likely related to treatment if statistically significant, but remaining within the range of historical controls.

• A difference in thyroid hormone levels is less likely to be adverse if there is no alteration in the general function of the thyroid. Therefore, the use of thyroid weights and histology may provide a better assessment of thyroid function (DeVito et al., 1999; Choksi et al., 2003). These endpoints are also less sensitive to confounders.

• A difference in thyroid hormone levels is less likely to be adverse if it is transient. Caloric restriction (St Germain and Galton, 1985; Cavalieri, 1991) influences thyroid hormone economy and may occur due to non-palatability of diets and reduced food and water intakes.

• A difference in thyroid hormone levels should not be considered as adverse if it is secondary to other adverse effects that occur at lower doses. Examples are chemicals that lower plasma thyroid hormone levels through interference with hormone transport, increase the hepatic uptake, hepatic metabolism and hepatic clearance of thyroid hormones, or enhance hepatic uptake and biliary clearance of thyroxine (Cunha and van Ravenzwaay, 2005). As a consequence, potential differences in thyroid hormone levels should not be considered as adverse when they occur only at dose levels that are higher than the overall systemic NOAEL.

Considerations along the same lines apply when assessing the potential adversity of relative thyroid weight changes in the

Table 2

Comparison of NOAELs for thyroid effects used in CAG grouping (Group 2B) regarding "follicular cells/the T3/T4 system" by EFSA (NOAELcag) with those used for deriving acceptable daily intake (NOAELADi) for the same pesticide based on selected examples.

Substance

ADIa NOAELadi NOAEL for thyroid [mg/kg bw] endpointb [mg/kg bw]

Possible rationale for difference in NOAEL settings

Fipronil 0.0002

Ioxynil 0.005

Desmedipham 0.03

Dithianon 0.01 1

0.003 0.3 0.04

Mepanipyrim 0.02 2

Cyhalofop-

butyl Pyrethrins

<0.02 <0.2 <3.86 <2.5 0.1 <6.6 <2.45

Decreased T3/T4; effect not considered adverse and not used for ADI setting Effects not considered adverse due to hepatic enzyme induction Decreased T4; effect not considered treatment-related and not used for ADI setting Decreased T3/T4; considered as isolated occurrence and not used for ADI setting

0.1 is based on follicular cell adenoma in rats, which are not mentioned in the EU Peer review document; ADI is based on liver toxicity in long-term mouse study

Decreased T3 levels, effect not considered for ADI-setting due to hepatic enzyme induction as presumed mechanism

Increased incidence of thyroid follicular carcinoma in 2-year rat study, effects not mentioned in EU Peer Review document; previously considered as the NOAEL; ADI is based on liver toxicity in rats.

a Taken from EC pesticides data base (http://ec.europa.eu/food/plant/pesticides/eu-pesticides-database/public. b Taken from EFSA-PPR-Panel, 2014.

absence of histopathological correlates.

2.2.2.2. Consequence for assignment to CAGs. Since the measurement of thyroid hormones is prone to confounders, thyroid hormones are not unambiguous biomarkers for adverse effects. As a consequence, a pesticide should only be assigned to the CAG thyroid group "follicular cells and T3/4" based on the following weight-of-evidence considerations:

• A clear dose-response is observed regarding changes in T3/T4 and TSH and the changes are consistent with the understanding of thyroid hormone homeostasis (T3/T4 Y and TSH [) (Cunha and van Ravenzwaay, 2005).

• The observed changes are outside the historical control ranges of the performing laboratory.

• A functional impairment attributable to the hormone changes has been observed (e.g., thyroid weight change) at higher doses or after longer exposures to the same dose levels.

• Absence of reversibility of changes in thyroid cell morphology/ number and in thyroid-pituitary hormones upon cessation of dosing is demonstrated.

• Changes in thyroid hormone levels alone need to be observed consistently in different studies considering the specific sensitivities of the animal model to thyroid effects.

If changes in thyroid parameters considered as adverse were identified in a single study, further evaluation of the consistency of effects is required to consider if the effect may qualify for inclusion of the pesticide in the CAG "thyroid follicular cells/the T3/T2 system". This requires consideration of all available information from toxicity studies using weight of evidence (WoE) approaches. Relevant aspects to be considered in such a WoE are consistency of results over studies integrating duration of study and dose levels were effects were observed, animal models and their specific sensitivity, and study quality.

2.2.3. Evaluation of the assignments of selected pesticides to the level 2 CAG "follicular cells/the T3/T4 system" based on these considerations

These criteria were applied to specific pesticides proposed to have thyroid effects due to the frequent finding of effects on the thyroid. From the 242 pesticides present in the database, EFSAs contractor concluded that 158 had effects on the thyroid system. Substances from different pesticide classes were selected for an in-depth review of available study reports with the aim to compare the results of a detailed analysis with those of EFSAs contractor assignment to the CAG "thyroid follicular cells/T3/T4 system".

2.2.3.1. Ametoctradin. Ametoctradin was identified by an EFSA contractor to induce specific effects on the thyroid system based on increased relative thyroid weights seen in an oral 90-day study in rats (RIVM-ICPS-ANSES, 2016). In this study performed according to OECD test guideline 408, groups of 10 male and 10 female Wistar rats received dietary concentrations of 0; 1500; 5000; or 15,000 ppm of Ametoctradin (corresponding to a mean daily dose of 0, 106/123, 358/416, or 1083/1235 mg/kg/day in male/female rats, respectively) (BASF, 2007). With respect to potential thyroid effects, evaluated parameters were limited to the determination of thyroid weights and pathology. While histopathology did not reveal treatment-related changes in the thyroid, small, but statistically significant increases in thyroid weight were observed at all doses in female animals (supplementary material, Table 2s).

Besides the thyroid weight changes, no other adverse effects were observed and the overall NOAEL was 15,000 ppm. Nevertheless, the isolated thyroid finding resulted in the conclusion that

Ametoctradin is a specific thyroid toxicant (RIVM-ICPS-ANSES, 2016). This conclusion is not supported when applying the criteria dose-response and database consistency. The minor increases in absolute and relative thyroid weight in female rats treated for 90 days have to be considered as incidental and not treatment-related due to the lack of a dose-response relationship, the absence of histopathological changes in the thyroid, and the fact that weight changes were limited to female animals. Furthermore, thyroid effects of Ametoctradin were only observed in this single study, the other repeated dose studies with Ametoctradin including two-year studies in rats and mice and three and 12-month dog studies did not indicate effects on the thyroid system (EFSA, 2012). Thus, based on an in-depth analysis of the original study report and a weight-of-evidence approach, the conclusion that Ametoctradin is a specific thyroid toxicant is not supported. In consequence, Ametoctradin should not be assigned to the CAG for thyroid effects.

2.2.3.2. Amidosulfuron. EFSA (2014a) also identified Amidosulfuron as having specific effects on the thyroid system based on follicular cell hypertrophy and follicular cell hyperplasia observed in an oral 28-day study in dogs. Groups of three male and three female Beagle dogs received dietary concentrations of 0, 400; 2000; or 10,000 ppm Amidosulfuron for one month (corresponding to a mean daily intake of 0, 26/24, 130/121, or 658/611 mg/kg/day in male/female dogs, respectively) (Bayer, 1988). With respect to potential thyroid effects, evaluated parameters were limited to the determination of thyroid weights and pathology. Although lacking statistical significance, a dose-related trend for increased absolute and relative thyroid weights was observed in mid and high-dose animals of both sexes. Relative thyroid weights were increased by 14/18% at 2000 ppm and by 26/27% at 10,000 ppm in male and female animals, respectively (supplementary material, Table 3s). In addition, histopathological examinations showed increased incidences in thyroid effects in mid- and high-dose dogs (supplementary material, Table 4s).

The overall NOAEL in this study was 400 ppm based on the histopathological findings in the thyroids. The highest dose level of 10,000 ppm produced signs of overt toxicity (frequent vomiting and precipitation of test substance in urine) and thus exceeded the maximum tolerated dose. The apparent dose-related trend in the increase in thyroid weight in mid and high dose dogs had histo-pathological correlates since follicular hypertrophy and follicular hyperplasia also increased in incidence and severity with dose. Thus, Amidosulfuron produced adverse affects on the thyroid system in this specific study consistent with EFSAs conclusion (EFSA, 2014b). However, additional studies in dogs at dose levels of up to 8000 ppm Amidosulfuron and daily intakes of up to 270 mg/kg/ day and for a much longer administration period of 52 weeks did not show effects on the thyroid system (Bayer, 1989, 1993). Likewise, no treatment-related findings in the thyroid were seen in several rodent studies with a duration of up to two years (EFSA-DAR, 2006). Thus, an in-depth analysis of all available study reports shows that the thyroid findings observed in a single 1-month study in dogs are spurious and may be related to the specific strain and age of the animals used in this study or other uncontrolled factors in the study. Therefore, an analysis of the overall available data does not indicate specific effects of Amidosulfuron on the thyroid and does not justify inclusion of Amidosulfuron into the CAG "thyroid follicular cells/T3/T4 system".

2.2.3.3. Cyhalofop-butyl. EFSA (2014a) identified Cyhalofop-butyl as having specific effects on the thyroid system based on thyroid follicular cell hypertrophy observed in oral 90-day and 52-week studies in dogs and due to follicular cell adenoma observed in a

lifetime study in rats. The 90-day repeat-dose toxicity study in dogs was performed according to OECD test guideline 409. Groups of four male and four female Beagle dogs received dietary concentrations of 0, 100; 500; or 2500 ppm for a period of 13 weeks (corresponding to a mean daily intake of 0, 2.9/3.2, 14.7/15.6, or 75.2/79.4 mg/kg/day in male/female dogs, respectively) (Dow, 1994b). With respect to potential thyroid effects, evaluated parameters were limited to the determination of thyroid weights and pathology. No toxicologically relevant effects on absolute or relative thyroid weights were observed. While a tendency for increased thyroid weights (statistically not significant) was seen in males at 500 and 2500 ppm (with absolute thyroid weights 20% and 18% higher in mid and high dose animals as compared to controls, respectively), there was no dose relationship, changes were limited to male animals, and the weight increase was not associated with histological changes. Thus, in agreement with the conclusions of the original study report (Dow, 1994b), this finding is not considered adverse. Histopathological examinations showed treatment-related findings only in the thyroids of high-dose females (follic-ular hypertrophy, incidence of 2/4). The overall NOAEL in this study was 100 ppm based on adverse effects on the liver and gallbladder at 500 ppm and higher.

To assess the relevance of the thyroid effects of Cyhalofop-butyl in dogs, a number of additional studies in dogs and in rodents are available. A 52-week study in dogs was performed according to OECD test guideline 452. Groups of four male and four female Beagle dogs received dietary concentrations of 0, 50; 300; or 1800 ppm for a period of 52 weeks (corresponding to a mean daily test substance intake of 0,1.2/1.3, 7.6/7.6, or 46.7/45.9 mg/kg/day in male/female dogs, respectively) (Dow, 1994a). Evaluated parameters were limited to the determination of thyroid weights and pathology. In male dogs, a dose-related increase in absolute and relative thyroid weights was observed. While this increase in male animals did not reach statistical significance, absolute thyroid weights in mid and high-dose animals were increased by 16 and 31% as compared to controls. At the same time, a depression of total body weight gain (29% lower as compared to controls) occurred in males receiving 300 or 1800 ppm and final body weight in those two dose groups were 12% lower than in control animals. In female animals, absolute (+18%) and relative thyroid (+14%) weight was only increased in high dose animals. In addition, histopathological examinations showed treatment-related findings in the thyroids of one high-dose female dog (follicular hypertrophy). The overall NOAEL was 50 ppm due to severe depression of body weight gain in males and adverse gallbladder effects in females at 300 ppm and higher. Besides the dog studies, the database on the animal toxicity of Cyhalfop-butyl also includes a number of rodent studies to evaluate the consistency of the thyroid findings. In mice, no findings in the thyroid system were recorded in 13-week and lifetime studies. In a combined chronic toxicity/carcinogenicity study, groups of 50 male and 50 female Fischer rats received dietary concentrations of 0, 3, 6, 24, or 100 ppm (males) or 0, 6, 60, or 600 ppm (females) for 104 weeks (corresponding to mean daily test substance intakes of 0, 0.10, 0.20, 0.82, or 3.44 mg/kg/day in males and 0.25, 2.48, or 24.97 mg/kg/day in females, respectively). Satellite groups of 40 males and 40 females per dosage level were included for interim sacrifices after 13, 26, 52, and 78 weeks (Dow, 1994c). With respect to potential thyroid effects, evaluated parameters were limited to organ pathology, organ weights were not recorded. Findings in the thyroids were limited to a not statistically significant increase in the incidence of neoplastic lesions (supplementary material, Table 5s). These findings are not considered as treatment-related as their incidence is within the range of their historical incidences. The mean incidence for follicular thyroid adenoma in male F344 rats is given as between 0.9% and 2.2%

(Tennekes et al., 2004). For thyroid C-cell carcinoma, historical incidences in female F344 rats are between 1.9% (Haseman et al., 1998) and 3.5% (Tennekes et al., 2004). The overall NOAEL in this study was 24 ppm due to adverse effects on the kidneys and liver.

With Cyhalofop-butyl, thyroid effects were consistently seen in the two dog studies. In male dogs, at doses of 300 ppm and higher, absolute and relative thyroid weights were increased without his-topathological correlate. In female dogs, thyroid weights were increased in high dose females after a treatment period of 52 weeks, and follicular hypertrophy was observed in animals treated with 1800 ppm and higher. In the absence of hormone measurements, thyroid weight increase is the most sensitive parameter for thyroid toxicity, even in the absence of histopathological findings (Capen, 1998; Klaassen and Hood, 2001; Cunha and van Ravenzwaay, 2005). The overall specific NOAEL for thyroid effects from the 13-week and 52-week dog studies can be set at 50 ppm. Therefore, an analysis of the overall available data indicates specific effects of Cyhalofop-butyl on the thyroid in the dog, but not in the rat. Thus, a weight-of-evidence approach does support an initial inclusion of Cyhalofop-butyl into the CAG "thyroid follicular cells/ T3/T4 system". Moreover, this NOAEL for thyroid effects is higher than the overall NOAEL used for ADI setting.

2.2.3.4. Dithianon. EFSA (EFSA-PPR-Panel, 2014) identified Dithia-non as having specific effects on the thyroid system based on a decrease of circulating T3 and/or T4-levels seen in a 4-week study in mice, a 90-day study in rats, a lifetime study in rats, and an increased relative thyroid weight and thyroid inflammation seen in a 52-week study in dogs. The database on Dithianon therefore apparently has consistent support for effects of this compound on the thyroid. Details of the studies are described below.

In a dose-range finding study, groups of six male and six female CD-1 mice received dietary concentrations of 0, 100; 500; or 1000 ppm for a period of four weeks (corresponding to mean daily test substance intakes of 0,15, 75, or 150 mg/kg/day) (EFSA-DAR, 2010). With respect to potential thyroid effects, evaluated parameters were limited to the determination of the levels of T3, T4, and TSH in plasma. Thyroid weights and thyroid histopathology were not assessed. Females of the mid and high dose groups showed a dose-related decrease in plasma T3 and T4 while males in the mid and high dose groups only showed a dose related decrease in plasma T4. No other effect on thyroid hormones was observed. The overall NOAEL in this study was 100 ppm based on thyroid hormone and liver and kidney changes. However, no effects on thyroid weight or pathology were seen in a 80-week carcinogenicity study in mice (EFSA-DAR, 2010) questioning the relevance of the thyroid findings from this short term study.

A number of rat studies also reported changes in thyroid hormone levels after Dithianon administration. In a 90-day study in rats performed according to OECD test guideline 408, groups of ten male and ten female Sprague-Dawley rats received dietary concentrations of 0, 30,180, and 1080 ppm of Dithianon for a period of 13 weeks (corresponding to mean daily intake of 0, 2.5/3.0, 14.6/ 16.3, or 86.7/99.5 mg/kg/day in male/female rats, respectively). An additional group of ten high-dose animals/sex was assigned to a four-week recovery period (BASF, 1989b). Evaluated thyroid parameters were plasma T3, T4, and TSH, and thyroid weights and histopathology. At all dose-levels, T3 concentrations were significantly lower as compared to controls in treated males, but without dose-response. In addition, the concentration of T4 in serum was significantly decreased in high dose male rats, but no changes in TSH were observed (supplementary material, Table 6s).

In the recovery group, T3 levels at the end of the four-week recovery period had returned to normal, but circulating T4 levels remained lower than in controls. No toxicologically relevant effect

on absolute and relative thyroid weight was observed, but the highest dose-level resulted in a significant reduction in body weight gain in both males and female rats (supplementary material, Table 7s). It should be noted that according to the study protocol, only the left thyroid was weighed which may limit the informative value of the data. Histopathology did not reveal treatment-related changes in the thyroids. The overall NOAEL in this study was 180 ppm due to adverse effects on body weight gain, renal pathology and liver, kidney and adrenal weights at the highest tested dose.

Besides the rodent studies, 90-day and 52-week studies in dogs are available for Dithianon. Groups of four male and four female Beagle dogs received dietary concentrations of 0, 40; 200; or 1000 ppm for 13 weeks corresponding to mean daily intake of 0, 0.6/0.7, 3.0, or 12.6 mg/kg/day in male/female dogs, respectively (BASF, 1989a). With respect to thyroid effects, evaluated parameters were plasma T3 and T4 levels, thyroid weights, and thyroid pathology. While statistically not significant, mean circulating T3 levels at the end of the treatment period were reduced in high dose male and female rats and circulating T4 levels may also be increased in a non-dose-related manner in females of the mid and high dose groups (supplementary material, Table 8s).

As the study did not determine TSH, the informative value of the tabulated data is somewhat diminished. No toxicologically relevant effect on absolute and relative thyroid weight was observed and the only histopathological observation was a non-dose related increased incidence of focal thyroiditis with squamous epithelial metaplasia in two out of four dogs at 200 ppm. The overall NOAEL in this study was 200 ppm due to adverse effects on body weight gain and kidney weights at the highest tested dose.

In the 52-week repeat-dose toxicity study with Dithianon, groups of four male and four female Beagle dogs received dietary concentrations of 0, 40; 200; or 1000 ppm corresponding to mean daily intake of 0,1.6, 7.3/7.9, or 37.1/37.5 mg/kg/day in male/female dogs (BASF, 1991). Evaluated thyroid parameters were thyroid weights and pathology. In high dose males and females, absolute and relative thyroid weights were increased as compared to controls, reaching statistical significance in the females (supplementary material, Table 9s). While the increase in male animals did not reach statistical significance, absolute (by 15%) and relative (by 23%) thyroid weights were increased. However, the observed weight changes are not considered to be treatment-related as they correlated with histopathological findings of species and strain-specific lymphocytic thyroiditis. A comparison of the individual organ weight data with the individual pathological findings leeds to the conclusion that the increase in mean absolute and relative thyroid weight for high dose animals can be attributed to the single male and the two female dogs diagnosed with lymphocytic thyroiditis. The overall NOAEL in this study was 40 ppm due to adverse effects on the liver and kidneys seen at 200 ppm and higher doses. Therefore, the original study report correctly concluded that the thyroid findings were not related to the administration of Dithia-non. For unknown reasons, the DAR concluded that Dithianon caused "thyroid toxicity".

For a further characterization of potential thyroid effects of Dithianon, a combined chronic toxicity/carcinogenicity study is available. Groups of 70 male and 70 female Sprague-Dawley rats received dietary concentrations of 0, 20, 120, or 600 ppm for 104 weeks corresponding to mean daily intake of 0, 1, 6, or 30 mg/kg/ day (EFSA-DAR, 2010). With respect to potential thyroid effects, evaluated parameters were plasma T3, T4, and TSH, thyroid weights, and thyroid histopathology. At study termination, statistically significant decreased T3 levels in high dose males were observed without changes in T4 and TSH (supplementary material, Table 10s) changes in absolute and relative thyroid weight were

observed. Histopathology did not reveal any treatment-related neoplastic or non-neoplastic changes in the thyroids. The overall NOAEL for chronic toxicity in this study was 20 ppm based on adverse effects on body weight gain and renal pathology observed at 120 ppm and higher.

In summary, Dithianon reduced levels of circulating T3 and T4 in rodents but did not influence TSH. Similarly, in dogs, reduced levels of circulating T3 were observed (TSH not determined). Dithianon did not induce thyroid pathology. The pattern of thyroid hormone changes without concomitant thyroid pathology is indicative of an indirect effect on the thyroid due to induction of thyroid UDP-glucuronosyltransferase. Some microsomal enzyme inducers do not increase serum TSH (Klaassen and Hood, 2001) and thus do not increase thyroid follicular cell proliferation. Hepatic enzyme induction as basis for the changes in thyroid hormones induced by Dithianon is supported by increased liver weights and hepatocel-lular hypertrophy in the 52-week dog study. In addition, the observed increases in absolute and relative thyroid weight in high dose animals seen at the end of a 52-week oral toxicity study in dogs correlated with lymphocytic thyroiditis (Gosselin et al., 1981). Lymphocytic thyroiditis is commonly found in colony-bred Beagle dogs and reported incidences in controls indicate that one in four dogs is affected (Benjamin et al., 1996). This conclusion also does not support a specific thyroid toxicity in this 52-week study by Dithianon. Thyroid findings seen in animal studies performed with Dithianon were either species-specific and not related to treatment, or not relevant for human risk assessment. Therefore, Dithianon should not be assigned to the CAG for thyroid effects.

3. Discussion

For cumulative risk assessment of pesticides in food, EFSA proposes a grouping based on target organ toxicity and to apply the concept of dose addition. Lacking data on mode of action, pesticides were allocated to CAGs based on a phenomenological approach, apparently using information on organ toxicity in the DARs. While the use of dose addition is appropriate but conservative (Teuschler, 2007; Meek et al., 2011), care needs to be taken in the process of the allocation of a pesticide to a specific group since the pesticide needs to induce clear and consistent adverse effects on the selected target organ in appropriately conducted toxicity studies (EFSA-PPR-Panel, 2014).,This requires a detailed assessment of the toxicity database on the specific pesticide integrating dose-response, consistency of effects over different studies, and reliance on clear adverse effects or an adverse outcome pathway that is well established. In addition, it needs to be assessed if toxic effects on the target system are the lead effects in the toxicity of the pesticide or are secondary to adverse effects on other target organs or biological responses such as enzyme induction. As this analysis shows, assignment of pesticides to CAGs may have widely diverse complexities that depend on the pesticide and its intended mode-of-action for toxicities, the nature of effects induced, and the target organ.

Assignment to CAGs is straightforward for organophosphate esters and carbamates that are designed for interaction with a specific molecular target, AChE. The interaction with AChE initiates the characteristic effects on the nervous system, requires specific structural features and inhibition of AChE can be readily determined with acceptable precision and little confounding in blood samples. AChE inhibition is also a more sensitive parameter than functional and pathological endpoints (US-EPA, 2006). Based on these considerations, allocation of chemicals to the level 2 CAG group "neurochemical effects" is straightforward and the allocations of individual pesticides to this group is well supported by structural features and by the biochemical marker AChE-inhibition. Nevertheless, appropriate studies in terms of exposure duration

and species investigated should be used to derive the NOAEL since steady state regarding AChE-inhibition is only reached after repeated exposures for more than three to four weeks (US-EPA, 2006).

In contrast, significant issues arise when pesticides need to be assigned to CAGs where the organ considered is not a target of intended pesticide toxicity, modes of action are poorly defined and likely multiple, and numerous confounders and inconsistent databases exist. While it is acknowledged that, generally, results from standard animal toxicity assays are adequate for assessing potential thyroid toxicity, a more detailed evaluation than that performed by the contractors to EFSA is needed to conclude on the presence of adverse effects. The detailed evaluation of the toxicity database on four pesticides assigned to the level 2 CAG "follicular cells and the T3/4 system" - relying on unambiguous adverse effects and considering the consistency of the information in the toxicity database - reveals that there is no scientifically valid support for inclusion in the level 2 CAG "thyroid follicular cells/the T3/T4 system" for at least three of the four examples. Reasons for this conclusion are the absence of dose-response for reported effects on thyroid function, the inconsistency of the database with higher quality studies or with longer exposure periods showing no effects on the selected target organ, and biologically inconsistent changes.

The four case studies presented above raise concern about the initial approach selected to assign pesticides to CAGs. The selection is apparently based on a cursory collection of the lowest NOAELs present in the database for a specific endpoint. Published DARs with their abbreviated study summaries do not provide adequate information to decide if observed effects are adverse and robust. Simply selecting the lowest effect level for an endpoint from an isolated study and not evaluating consistency of the overall toxicity database is not considered good toxicological practice and inconsistent with the well justified recommendations of EFSA regarding the basis of grouping into CAGs (EFSA-PPR-Panel, 2014). Therefore, to assign to the level 2 CAG "follicular cells/the T3/T4 system", the following points need to be evaluated in detail for each pesticide:

• Confirm and validate findings of thyroid toxicity stated in DAR based on an analysis of the original study report

• Consider consistency of the complete set of repeat-dose toxicity studies

• Consider the difference between its critical NOAEL (used in the regulatory context for reference value setting) and its NOAEL for a specific effect: "For instance, the contribution of a pesticide to the cumulative toxicity in a CAG will be inversely proportional to the difference in values between its NOAEL for the specific effect and its critical NOAEL. That is, if the difference between these NOAELs is small the contribution of a substance to the cumulative toxicity in a CAG will be higher than the contribution of a substance for which the difference between its NOAELs is large." (EFSA-PPR-Panel, 2014)

As indicated by the detailed analysis of some selected pesticides, the generic approach used by EFSAs contractors is inadequate for a correct hazard-characterization and, thus, for the assignment of substances into their respective CAGs. It is expected that a detailed analysis of the toxicity database for other chemicals allocated to the CAG "Toxicity to the thyroid hormone system (gland or hormones" will reveal more examples whose inclusion is insufficiently supported. An initial screening of the overall database may therefore only serve as a first step to identify potential candidate pesticides for a CAG. The final assignment has to be based on an assessment following established criteria as outlined (EFSA-PPR-Panel, 2014). Apparently, EFSA has not yet decided on detailed approaches regarding the cumulative assessment of pesticides included in a

CAG and how to derive reference values since "reference values for the specific effects within the different CAGs and index compounds (IC) may be established, but such considerations are outside the scope of the present Opinion on developing a general methodology. This would represent a future step and would require the identification of IC within each CAG" (EFSA-PPR-Panel, 2013b, 2014). For cumulative risk assessment, relative potency data for all members of the CAG are required to obtain a well-supported cumulative assessment. While there seems to be no apparent guidance to define relative potency, the contractors to EFSA have assembled NO(A)ELs derived from studies with different species, duration and dosing with a focus on the lowest effect concentrations. The use of the lowest NO(A)EL from any study available to derive relative potency factors has a number of significant weaknesses:

- Focus on the NO(A)EL from any study will induce a considerable uncertainty in the relative potency factors due to differing study designs and animal models that may show widely differing susceptibilities

- Study duration and study quality are not considered

- The use of NO(A)ELs does not consider dose-response and dosespacing in a study and studies without a NO(A)EL cannot be included in the assessment without additional extrapolation factors that induce further uncertainty

To derive relative potency factors, it is therefore recommended to rely on studies in one species based on designs widely applied in the regulatory testing requirements for pesticides. For example, the US EPA used AChE-inhibition data in brain obtained from 90-day oral toxicity studies in female rats with the pesticide of interest to derive relative potency factors for the endpoint AChE inhibition. Such an approach relies on datasets that can be well compared and are readily available. Results from studies with longer duration are required specifically for responses with complex time-effect relations ensuring that steady state is reached. EPA also uses brain AChE-activity as an endpoint since the data are less variable and closer to the actual adverse effect of relevance as compared to blood AChE (US-EPA, 2006). Use of comparative datasets ensures that confounding is minimized and comparative potency is precise. Results from standardized 90-day oral toxicity studies are readily available for most registered pesticides. Moreover, when relying on such studies, benchmark doses can be calculated with acceptable precision due to the larger numbers of animals/dose group and at least three dose groups. The use of benchmark doses eliminates most of the issues regarding inadequate dose-spacing or absence of a NOAEL and provides a mathematically meaningful numerical value for comparative potency. Relying on a study with standard design and sufficient animals/dose group will also avoid the issues of low statistical power from studies with few animals/dose group such as most dog studies that were used to derive NOAELs/LOAELs.

Calculation of benchmark doses can also be applied to derive comparative potency data for members of the "thyroid follicular cells/T3/T4 system" CAG using common endpoints such as consistent thyroid weight changes and histopathology that are available from most studies. This will also avoid reliance on potentially spurious data from studies with low numbers of animals/group.

It needs to be recognized that the calculation of relative potencies is not related to the issue of the most sensitive species to be used to derive the POD, but has to rely on the best available dataset for one species. Issues of species extrapolation are more appropriately addressed in the overall extrapolation factor applied to the summary value for the CAG. Normally, a margin-of-exposure (MoE) of >100 is considered of low concern and may also be applied to critical effects regarding CAGs. Based on a detailed evaluation of the database of many pesticides assigned to the CAG "neurochemical

effects" due to available comparative data on the biomarker AChE-inhibition, the US EPA concluded that the standard MoE >100 is sufficient for most pesticides assigned to this group to protect against effects even in potentially sensitive subgroups. Therefore, for reasons of resource conservation, this evaluation could be accepted by EFSA for their conclusions and it is recommended that a MoE >100 based on a BMD10 should be used for this CAG.

However, a MoE of >100 may not be required for "thyroid follicular cells/T3/T4 system" CAG due to well known species differences in thyroid function between rodents and humans (Table 3). While the basic thyroid function is comparable between rodents and humans, there are significant differences with respect to hormone transport in blood and metabolic clearance of thyroid hormones. The binding affinity of TBG for T4 is approximately 1000 times higher than for albumin and the percentage of unbound T4 is significantly higher in rodents. As a consequence, when thyroid function is impaired, rodents need about 10 times more T4 (20 mg/ kg body weight) for full substitution as compared to an adult human (2.2 mg/kg body weight). The lower percentage of protein binding also results in significant shorter plasma half-lives for T3 and T4 in rats (12—24 h for T4 and 6 h for T3) as compared to humans (five - nine days for T4 and one day for T3). In addition, there is a marked sex-difference in serum TSH levels in rodents, but not in primates. Given these differences, rodents are more sensitive to disturbances of the HPT-axis and more susceptible for the development of thyroid hyperplasia and/or tumors as a consequence of chronic TSH stimulation (Capen, 1998; Lewandowski et al., 2004; Wu and Farrelly, 2006). Therefore, caution must be applied when extrapolating adverse thyroid findings seen in rodents to humans and the interpretation of thyroid hormone levels needs to consider the marked species differences. In humans, abnormal thyroid hormone measurements in patients without impairment of the thyroid are rarely predictive of an underlying disease. While up to 70% of those patients show significant changes of T3, T4 or TSH, the incidence of true hypothyroidism or hyperthyroidism is less than 1% (Cavalieri, 1991; Davies and Franklyn, 1991). Chemicals that permanently increased TSH levels have been shown to induce thyroid tumors in rodents (Capen, 1997; Hurley, 1998; Hood et al., 1999). However, in humans ionizing radiation is the only established human thyroid carcinogen (Hard, 1998). Strikingly, sulfonamides, which have been used in therapy for over 70 years in comparatively high oral dosages (grams per day), produce, at best, only mild changes in thyroid function and no evidence of thyroid tumors in humans, while they cause neoplastic and preneoplastic changes in the thyroid glands of experimental animals (Capen, 1997; Poirier et al., 1999). Furthermore, rodents have been shown to be more susceptible to some secondary

mechanisms for thyroid toxicity as compared to humans. For example, substances that interfere with hormone transport carriers will not reduce both T3 and T4 in humans due to the high-affinity serum carrier protein thyroxine-binding globulin that exists in humans. Histamine antagonists have been shown to decrease T4 and reduce TSH in rats by increasing the hepatic clearance of thyroid hormones through increased hepatic accumulation. No such effect was seen in mice, dogs and humans (Cunha and van Ravenzwaay, 2005). Many environmental chemicals and drugs induce liver microsomal enzymes resulting in an increase in T4 serum clearance and/or increase in T4 UDP-glucuronosyl-transferase activity. Due to the greater proportion of free thyroid hormone in serum, rodents are more sensitive than humans to this mode of action (Curran and DeGroot, 1991; Capen, 1997; Hurley, 1998).

It is also well understood that significant disruptions in thyroid function during pregnancy can cause neurological deficits in offspring. In humans, functionally significant changes in T4 production during development are known to be associated with developmental delay, low body mass, brain developmental abnormalities and neurobehavioral developmental disorders (Di Liegro, 2008; Hartoft-Nielsen et al., 2011; Negro and Mestman, 2011; Koibuchi, 2013; Forhead and Fowden, 2014). However, potential neurodevelopmental effects in children only occur after very marked maternal hypothyroidism or hypothyroxinemia. Significant decreases in T4 levels (>40%) during gestation and the early postnatal period did not alter synaptic transmission in the dentate gy-rus of the hippocampus of rat pups (Gilbert, 2003). In rats, moderate maternal hypothyroxinemia during gestation does not necessarily lead to neurodevelopmental behavioral deficits in the rat offspring (Axelstad et al., 2011a, 2011b). While thyroid hormone levels alone seem to be poor predictors of potential neuro-developmental effects of thyroid hormone disrupting compounds, observed neurodevelopmental deficits seen after exposure to pro-pylthiouracil were accompanied or preceded by significant effects on the thyroid weight and histology, and decreases in litter size, pup survival, and pup body weight gain (Brosvic et al., 2002; Noda et al., 2005).

In conclusion, the level 2 CAG "thyroid follicular cells/T3/T4 system" requires a much lower MoE as compared to other toxicity endpoints if based on rat data. It is recommended to apply a minimum MoE of 10 and rely on relative potency data based on BMD1o for increases in thyroid weight from 90-day oral rat studies for this CAG. Generally, a much more detailed assessment of toxicity data is required to provide a scientifically sound risk characterization of pesticide mixtures and a summary of the proposed approach is outlined in Fig. 1.

Table 3

Comparison of key aspects of the thyroid system between humans and rodents, from (Jahnke et al., 2004).

Parameter Human Rat Mouse

Half-life of T4 5—9 days 0.5—1 days 0.5—0.75 days

Half-life of T3 1 day 0.25 days 0.45 days

High-affinity thyroxine-binding globulin Present Absent Absent

Primary serum-binding protein Thyroxin-binding globulin Albumin Albumin

Sex differences in serum TSH levels No difference Adult males > adult females Adult males > adult

females

Amount of T4 supplementation required in 2.2 mg/kg bw/day 20 mg/kg bw/day Information not

absence of functional thyroid gland available

Development of the fetal pituitary—thyroid TSH and T3 secretion increases between Thyroid hormone synthesis and TSH- Information not

axis 18 and 20 weeks gestation secreting cells appear by gestation day available

Becomes functional late in the 1st and 17

early 2nd trimester Becomes functional in late gestation

Maturation appears complete by 4th and postnatally

postnatal week Maturation appears complete by 4th

postnatal week

Identify the key (most potent) compound

Use BMD10s as relative potency factors

Fig. 1. Summary of proposed assessment of relative potency for the level 2 CAG "thyroid follicular cells and the T3/4 system".

Acknowledgements

Drs. Dekant and Colnot were financially supported by ECPA to develop this manuscript. The views expressed in this manuscript are based on the professional judgment of the authors and do not necessarily represent those of ECPA.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.yrtph.2016.12.004.

Transparency document

Transparency document related to this article can be found online at http://dx.doi.org/10.1016Zj.yrtph.2016.12.004.

References

Axelstad, M., Boberg, J., Hougaard, K.S., Christiansen, S., Jacobsen, P.R., Mandrup, K.R., Nellemann, C., Lund, S.P., Hass, U., 2011a. Effects of pre- and postnatal exposure to the UV-filter octyl methoxycinnamate (OMC) on the reproductive, auditory and neurological development of rat offspring. Toxicol. Appl. Pharmacol. 250, 278—290. Axelstad, M., Boberg, J., Nellemann, C., Kiersgaard, M., Jacobsen, P.R., Christiansen, S., Hougaard, K.S., Hass, U., 2011b. Exposure to the widely used fungicide man-cozeb causes thyroid hormone disruption in rat dams but no behavioral effects in the offspring. Toxicol. Sci. 120, 439—446. Ballantyne, B., Marrs, T., Syverson, T. (Eds.), 2009. General and Applied Toxicology. Wiley, New York.

BASF, 1989a. 90-Day Feeding Study in Beagle Dogs. BASF AG, Ludwigshafen, Germany.

BASF, 1989b. 90-Day Feeding Study in Sprague-Dawley Rats. BASF AG, Ludwigshafen, Germany.

BASF, 1991. 2 Week Oral (Dietary Administration) Toxicity Stuy in the Beagle. BASF

AG, Ludwigshafen, Germany. BASF, 2007. BASF 650 F - Repeated Dose 90-day Oral Toxicity in Wistar Rats; Administration in the Diet (MRID#: 47700051) Performed by Laboratory of Experimental Toxicology and Ecotoxicology. BASF AG, Ludwigshafen, Germany. Report No. 50S0189/04066; sponsored by BASF Corporation, Research Triangle Park, NC, USA.

Bayer, 1988. Testing for Toxicity by Repeated Oral Administration to Beagle Dogs for

1 Month (Range-finding-test). Bayer SAS, Valbonne, France. Bayer, 1989. Testing for Toxicity by Repeated Oral Administration to Beagle Dogs (3-

month Feeding Study). Bayer SAS, Valbonne, France. Bayer, 1993.52-Week Oral Toxicity (Feeding) Study in the Dog. Bayer SAS, Valbonne, France.

Benjamin, S.A., Stephens, L.C., Hamilton, B.F., Saunders, W.J., Lee, A.C., Angleton, G.M., Mallinckrodt, C.H., 1996. Associations between lymphocytic thyroiditis, hypothyroidism, and thyroid neoplasia in beagles. Veterinary Pathol. Online 33, 486—496.

Brosvic, G.M., Taylor, J.N., Dihoff, R.E., 2002. Influences of early thyroid hormone manipulations: delays in pup motor and exploratory behavior are evident in

adult operant performance. Physiol. Behav. 75, 697—715.

Capen, C.C., 1997. Mechanistic data and risk assessment of selected toxic end points of the thyroid gland. Toxicol. Pathol. 25, 39—48.

Capen, C.C., 1998. Correlation of mechanistic data and histopathology in the evaluation of selected toxic endpoints of the endocrine system. Toxicol. Lett. 102—103, 405—409.

Carmichael, N., Bausen, M., Boobis, A.R., Cohen, S.M., Embry, M., Fruijtier-Polloth, C., Greim, H., Lewis, R., Bette Meek, M.E., Mellor, H., Vickers, C., Doe, J., 2011. Using mode of action information to improve regulatory decision-making: an ECE-TOC/ILSI RF/HESI workshop overview. Crit. Rev. Toxicol. 41,175—186.

Cavalieri, R.R., 1991. The effects of nonthyroid disease and drugs on thyroid function tests. Med. Clin. North Am. 75, 27—39.

Chambers, J.E., Meek, E.C., Chambers, H.W., 2010. The metabolism of organophos-phorus insecticides. In: Krieger, R. (Ed.), Hayes' Handbook of Pesticide Toxicology. Academic Press, San Diego, pp. 1399—1407.

Choksi, N.Y., Jahnke, G.D., St Hilaire, C., Shelby, M., 2003. Role of thyroid hormones in human and laboratory animal reproductive health. Birth Defects Res. B Dev. Reprod. Toxicol. 68, 479—491.

Cunha, G.C., van Ravenzwaay, B., 2005. Evaluation of mechanisms inducing thyroid toxicity and the ability of the enhanced OECD Test Guideline 407 to detect these changes. Arch. Toxicol. 79, 390—405.

Curran, P.G., DeGroot, L.J., 1991. The effect of hepatic enzyme-inducing drugs on thyroid hormones and the thyroid gland. Endocr. Rev. 12,135—150.

Davies, P.H., Franklyn, J.A., 1991. The effects of drugs on tests of thyroid function. Eur. J. Clin. Pharmacol. 40, 439—451.

Dekant, W., Bridges, J., 2016. A quantitative weight of evidence methodology for the assessment of reproductive and developmental toxicity and its application for classification and labeling of chemicals. Regul. Toxicol. Pharmacol. 82,173—185.

DeVito, M., Biegel, L., Brouwer, A., Brown, S., Brucker-Davis, F., Cheek, A.O., Christensen, R., Colborn, T., Cooke, P., Crissman, J., Crofton, K., Doerge, D., Gray, E., Hauser, P., Hurley, P., Kohn, M., Lazar, J., McMaster, S., McClain, M., McConnell, E., Meier, C., Miller, R., Tietge, J., Tyl, R., 1999. Screening methods for thyroid hormone disruptors. Environ. Health Perspect. 107, 407—415.

Di Liegro, I., 2008. Thyroid hormones and the central nervous system of mammals (Review). Mol. Med. Rep. 1, 279—295.

Dietrich, J.W., Landgrafe, G., Fotiadou, E.H., 2012. TSH and thyrotropic agonists: key actors in thyroid homeostasis. J. Thyroid. Res. 2012, 351864.

Dohler, K.D., Gartner, K., von zur Muhlen, A., Dohler, U., 1977. Activation of anterior pituitary, thyroid and adrenal gland in rats after disturbance stress. Acta Endocrinol. (Copenh) 86, 489—49 .

Dow, 1994a. 2-Month Oral Chronic Toxicity Study in Dogs. Dow Agrosciences Ltd, Abingdon, UK.

Dow, 1994b. 3-Week Oral Subchronic Toxicity Study in Dogs. Dow Agrosciences Ltd, Abingdon, UK.

Dow, 1994c. 24-Month Oral Chronic Toxicity and Oncogenicity Study in Rats. Dow Agrosciences Ltd, Abingdon, UK.

EC-SCHER, 2012. SCCS, SCENIHR, Opinion on the Toxicity and Assessment of Chemical Mixtures. Available online at: http://ec.europa.eu/health/scientific_ committees/environmental_risks/docs/scher_o_155.pdf.

EFSA, 2012. (European Food Safety Authority) Conclusion on the peer review of the pesticide risk assessment of the active substance ametoctradin (BAS 650 F). EFSAJ. 10 (n/a-n/a).

EFSA, 2013. (European Food Safety Authority) Conclusion on the peer review of the

pesticide risk assessment of the active substance fluopyram. EfSAJ. 11,3052 [76 pp.].

EFSA, 2014a. (European Food Safety Authority) Outcome of the public consultation on the Scientific Opinion on the identification of pesticides to be included in cumulative assessment groups (CAGs) on the basis of their toxicological profile. EFSA Support. Publ. 11, 53.

EFSA, 2014b. (European food safety authority) reasoned opinion on the review of the existing maximum residue levels (MRLs) for amidosulfuron according to article 12 of regulation (ec) No 396/2005. EFSA J. 12 (n/a-n/a).

EFSA-DAR, 2006. Draft Assessment Report - Public Version. Initial Risk Assessment provided by the Rapporteur Member State Austria for the Existing Active Substance AMIDOSULFURON of the Third Stage (Part a) of the Review Programme Referred to in Article 8(2) of Council Directive 91/414/EEC (March 2006).

EFSA-DAR, 2010. Additional Report to Draft Assessment Report Prepared in the Context of the Possible Inclusion of the Acrive Substance DITHIANON in Annex I of the of Council Directive 91/414/EEC provided by the Rapporteur Member State Hellas (January 2010).

EFSA-PPR-Panel, 2013a. (EFSA Panel on Plant Protection Products and their Residues) Scientific Opinion on the identification of pesticides to be included in cumulative assessment groups on the basis of their toxicological profile. EFSA J. 11. 3293-n/a.

EFSA-PPR-Panel, 2013b. (EFSA Panel on Plant Protection Products and their Residues). Scientific Opinion on relevance of dissimilar mode of action and its appropriate application for cumulative risk assessment of pesticides residues in food. EFSA J. 11 (3472), 3440.

EFSA-PPR-Panel, 2014. (EFSA Panel on Plant Protection Products and their Residues) Scientific Opinion on the identification of pesticides to be included in cumulative assessment groups on the basis of their toxicological profile (2014 update). EFSA J. 11,131.

Forhead, A.J., Fowden, A.L., 2014. Thyroid hormones in fetal growth and prepartum maturation. J. Endocrinol. 221, R87—R103.

Gartner, K., Buttner, D., Dohler, K., Friedel, R., Lindena, J., Trautschold, I., 1980. Stress response of rats to handling and experimental procedures. Lab. Anim. 14, 267—274.

St Germain, D.L., Galton, V.A., 1985. Comparative study of pituitary-thyroid hormone economy in fasting and hypothyroid rats. J. Clin. Invest 75, 679—688.

Gilbert, M.E., 2003. Perinatal exposure to polychlorinated biphenyls alters excitatory synaptic transmission and short-term plasticity in the hippocampus of the adult rat. Neurotoxicology 24, 851 —860.

Gosselin, S.J., Capen, C.C., Martin, S.L., 1981. Histologic and ultrastructural evaluation of thyroid lesions associated with hypothyroidism in dogs. Vet. Pathol. 18, 299—309.

Hard, G.C., 1998. Recent developments in the investigation of thyroid regulation and thyroid carcinogenesis. Environ. Health Perspect. 106, 427—436.

Hartoft-Nielsen, M.L., Boas, M., Bliddal, S., Rasmussen, A.K., Main, K., Feldt-Rasmussen, U., 2011. Do thyroid disrupting chemicals influence foetal development during pregnancy? J. Thyroid. Res. 2011, 342189.

Haseman, J.K., Hailey, J.R., Morris, R.W., 1998. Spontaneous neoplasm incidences in Fischer 344 rats and B6C3F1 mice in two-year carcinogenicity studies: a National Toxicology Program update. Toxicol. Pathol. 26, 428—44 .

Hood, A., Liu, Y.P., Gattone 2nd, V.H., Klaassen, C.D., 1999. Sensitivity of thyroid gland growth to thyroid stimulating hormone (TSH) in rats treated with anti-thyroid drugs. Toxicol. Sci. 49, 263—271.

Hoyer, P.B., Flaws, J., 2013. Toxic responses of the endocrine system. In: Klaassen, C.D. (Ed.), Casarett and Doull's Toxicology. The Basic Science of Poisons. McGraw-Hill Medical Publishing Division, New York, pp. 907—930.

Hurley, P.M., 1998. Mode of carcinogenic action of pesticides inducing thyroid follicular cell tumors in rodents. Environ. Health Perspect. 106, 437—445.

Jahnke, G.D., Choksi, N.Y., Moore, J.A., Shelby, M.D., 2004. Thyroid toxicants: assessing reproductive health effects. Environ. Health Perspect. 112, 363—368.

Kienzler, A., Bopp, S.K., van der Linden, S., Berggren, E., Worth, A., 2016. Regulatory assessment of chemical mixtures: requirements, current approaches and future perspectives. Regul. Toxicol. Pharmacol. 80, 321—334.

Klaassen, C.D. (Ed.), 2013. Casarett and Doull's Toxicology. The Basic Science of Poisons. McGraw-Hill Medical Publishing Division, New York.

Klaassen, C.D., Hood, A.M., 2001. Effects of microsomal enzyme inducers on thyroid

follicular cell proliferation and thyroid hormone metabolism. Toxicol. Pathol. 29, 34—40.

Klimisch, H.J., Andreae, M., Tillmann, U., 1997. A systematic approach for evaluating the quality of experimental toxicological and ecotoxicological data. Regul. Toxicol. Pharmacol. 25, 1 —5.

Koibuchi, N., 2013. The role of thyroid hormone on functional organization in the cerebellum. Cerebellum 12, 304—306.

Lamb, J.C.t., Boffetta, P., Foster, W.G., Goodman, J.E., Hentz, K.L., Rhomberg, L.R., Staveley, J., Swaen, G., Van Der Kraak, G., Williams, A.L., 2015. Comments on the opinions published by Bergman, et al. (2015) on Critical Comments on the WHO-UNEP State of the Science of Endocrine Disrupting Chemicals (Lamb et al., 2014). Regul. Toxicol. Pharmacol. 73, 754—757.

Lewandowski, T.A., Seeley, M.R., Beck, B.D., 2004. Interspecies differences in susceptibility to perturbation of thyroid homeostasis: a case study with perchlo-rate. Regul. Toxicol. Pharmacol. 39, 348—362.

Lewis, R.W., Billington, R., Debryune, E., Gamer, A., Lang, B., Carpanini, F., 2002. Recognition of adverse and nonadverse effects in toxicity studies. Toxicol. Pathol. 30, 66—74.

Lutter, R., Abbott, L., Becker, R., Borgert, C., Bradley, A., Charnley, G., Dudley, S., Felsot, A., Golden, N., Gray, G., Juberg, D., Mitchell, M., Rachman, N., Rhomberg, L., Solomon, K., Sundlof, S., Willett, K., 2015. Improving weight of evidence approaches to chemical evaluations. Risk Anal. 35,186—192.

Meek, M.E., Boobis, A.R., Crofton, K.M., Heinemeyer, G., Raaij, M.V., Vickers, C., 2011. Risk assessment of combined exposure to multiple chemicals: a WHO/IPCS framework. Regul. Toxicol. Pharmacol. 60, S1—S14.

Negro, R., Mestman, J.H., 2011. Thyroid disease in pregnancy. Best. Pract. Res. Clin. Endocrinol. Metab. 25, 927—943.

Noda, S., Muroi, T., Takakura, S., Sakamoto, S., Takatsuki, M., Yamasaki, K., Tateyama, S., Yamaguchi, R., 2005. Preliminary evaluation of an in utero-lactation assay using 6-n-propyl-2-thiouracil. Arch. Toxicol. 79, 414—421.

Poirier, L.A., Doerge, D.R., Gaylor, D.W., Miller, M.A., Lorentzen, R.J., Casciano, D.A., Kadlubar, F.F., Schwetz, B.A., 1999. An FDA review of sulfamethazine toxicity. Regul. Toxicol. Pharmacol. 30, 217—222.

RIVM-ICPS-ANSES, 2016. Toxicological Data Collection and Analysis to Support Grouping of Pesticide Active Substances for Cumulative Risk Assessment of Effects on the Nervous System, Liver, Adrenal, Eye, Reproduction and Development and Thyroid System (GP/EFSA/PRAS/2013/02). EFSA Supporting Publications, p. 184, 13, EN999.

Simon, T.W., Simons Jr., S.S., Preston, R.J., Boobis, A.R., Cohen, S.M., Doerrer, N.G., Fenner-Crisp, P.A., McMullin, T.S., McQueen, C.A., Rowlands, J.C., 2014. The use of mode of action information in risk assessment: quantitative key events/dose-response framework for modeling the dose-response for key events. Crit. Rev. Toxicol. 44 (Suppl 3), 17—43.

Sturla, S.J., Boobis, A.R., FitzGerald, R.E., Hoeng, J., Kavlock, R.J., Schirmer, K., Whelan, M., Wilks, M.F., Peitsch, M.C., 2014. Systems toxicology: from basic research to risk assessment. Chem. Res. Toxicol. 27, 314—329.

Tennekes, H., Kaufmann, W., Dammann, M., van Ravenzwaay, B., 2004. The stability of historical control data for common neoplasms in laboratory rats and the implications for carcinogenic risk assessment. Regul. Toxicol. Pharmacol. 40, 293—304.

Teuschler, L.K., 2007. Deciding which chemical mixtures risk assessment methods work best for what mixtures. Toxicol. Appl. Pharmacol. 223,139—147.

US-EPA, 2006. Environmental Protection Agency, Office of Pesticide Programs. Organophosphorus Cumulative Risk Assessment, 2006 Update.

WHO/IPCS, 2004. Exposure Assessment and Risk Assessment Terminology. Available at: http://www.who.int/ipcs/methods/harmonization/areas/terminology/ en/index.html [Last accessed February 2015].

Wu, K.M., Farrelly, J.G., 2006. Preclinical development of new drugs that enhance thyroid hormone metabolism and clearance: inadequacy of using rats as an animal model for predicting human risks in an IND and NDA. Am. J. Ther. 13, 141—144.