The World Health Organization defines an Endocrine Disruptor (ED) as a chemical substance or mixture that alters function of the endocrine system and consequently causes adverse health effects in intact organisms or their progeny or sub-populations.¹  This widely accepted definition of an ED consists of three elements: (1) the substance has to cause adverse health effects in intact organisms; (2) it has to alter the function of the endocrine system and (3) (1) has to be a consequence of (2). Consequently, to identify an ED by toxicity testing (1) it has to be shown that an adverse effect would occur in vivo, observable in a test animal system, epidemiologically or clinically;²  (2) it has to be demonstrated that the mechanism by which the substance is causing the adverse effect is endocrine and (3) a plausible link between the adverse effect and the endocrine mechanism has to be established. This definition and its elements have been widely accepted³  and are the basis for the ED criteria set by the European Union.^4,5  Since there are several elements required to fulfill the definition no single test can demonstrate whether or not a substance is an ED.

For detection of the adverse effect required under (1), it is important to note that the term “intact organism” does not necessarily mean that the adverse effect has to be demonstrated in an intact test animal. Adequately validated alternative test systems that are predictive of adverse effects in humans could also be used.⁵ 

As there is a lack of alternative test systems that fulfill these criteria, one or more appropriate in vivo studies, for example toxicity tests conducted according to OECD Test Guidelines, are still required to detect an adverse effect (1). In contrast, the endocrine mechanism requested under (2) can be detected in vitro, e.g. by a receptor binding or transactivation assay also within a screening program.

These two elements may be connected either by a top-down procedure or by a bottom-up procedure: For data-rich substances like pesticide active substances or high production volume chemicals, where several in vivo tests are required by the respective regulations, relevant effects would be likely to be detected first from the results of regulatory in vivo testing. Based on positive findings in vivo, mechanistic in vitro assays would be requested on a case by case basis to investigate if an endocrine mechanism is the cause for the observed effect in vivo. This procedure is in line with recent EU legislation and the European Chemicals Agency (ECHA) and the European Food Safety Authority (EFSA) guidance document.⁶  Since this top-down procedure starts with the apical in vivo studies, only specific mechanistic tests will be conducted based on positive findings and no screening is necessary. This situation is different for data-poor substances like low production volume or orphan chemicals (i.e. substances that do not have a producer any more but are still in the environment). Here no or very limited in vivo data is available. Hence, a bottom-up procedure starting with a number of in vitro assays to address different potential mechanisms by which the endocrine system could be disturbed would be a better starting point to avoid animal testing. Based on findings obtained from screening assays, in vitro substances would be prioritized for further testing in vivo. This bottom-up procedure was also chosen by the Endocrine Disruptor Screening Program (EDSP) of the US Environmental Protection Agency (EPA). However, the complexity of the endocrine system, the various mechanisms by which an ED may interfere with the endocrine system and limitations in the availability of standardized mechanistic assays still leave room for improvement.

As shown in Figure 1.1 the endocrine system consists of several organs including the pituitary, thyroid, adrenals, ovaries, testes, pancreas and many others. The endocrine system is organized in several axes that consist of hormone-mediated signaling cascades and feedback loops. One such axis is the hypothalamic–pituitary–gonadal (HPG) axis, another one the hypothalamic–pituitary–thyroid (HPT) axis. Since early research on endocrine disruption has focused on substances interfering with reproduction, development of assays has focused on the HPG axis in the past. This has now been extended to the HPT axis in recent EU programs and will be extended further in the future. Figure 1.1 shows different parts of the endocrine system and highlights the availability of standardized assays.

EDs can interfere with the endocrine system by several mechanisms: they may directly interact with receptors responsible for hormonal signaling such as the androgen receptor (AR), the estrogen receptors (ERα or ERβ) or the thyroid receptors (TRs). They may also affect the metabolism of hormones, either by inhibiting the biosynthesis of specific hormones like estradiol or by inhibiting or accelerating the degradation of hormones. In addition, they may also interfere with transport of hormones, their storage or release from glandular cells or with feedback loops or signaling cascades responsible for hormone action in cells. Figure 1.2 illustrates various mechanisms of ED action.

An example of a substance interfering with a hormone receptor is vinclozolin. This pesticide active substance causes inhibition of AR binding. The male sex hormone testosterone normally activates AR. This leads to alterations in gene expression, which consequently affects the function of cells of androgen responsive tissues in vivo. By inhibiting the AR, vinclozolin and similar substances reveal anti-androgenic properties. This means, for example, that in rats treated with vinclozolin a lower weight of male sex organs is observed.

To detect such mechanisms, a number of assays have been developed that are capable of detecting hormone-receptor interactions and respective (ant)agonistic properties, for example the androgen receptor transactivation assay (ARTA, OECD TG 458) or the estrogen receptor transactivation assay (ERTA, OECD TG 455).

Additional assays are capable of detecting interference with the biosynthesis of steroid hormones, e.g. the H295R steroidogenesis assay (OECD TG 456). H295R is a human adrenal cell line that is capable of synthesizing a number of steroid hormones like estradiol, cortisol, testosterone and progesterone. When these cells are treated with a substance capable of inhibiting one (or more) enzyme(s) important for the biosynthesis of one (or more) of these steroids, the concentration of the respective hormone is altered. This can also be exploited for the development of medicinal drugs.

Many of the enzymes important in biosynthesis of steroid hormones are cytochrome-P-450-(CYP)-enzymes. Letrozol, for example, is a triazole that is specifically designed to inhibit CYP19A1, the aromatase responsible for the biosynthesis of estradiol. Letrozol significantly reduces the concentration of estradiol in vitro and in vivo and is therefore used as a drug in estradiol-dependent breast cancer. However, substances that are actually intended to inhibit other CYPs like e.g. triazole fungicides targeting a fungal CYP, frequently also inhibit steroid hormone biosynthesis in mammals.

Serum concentration of the thyroid hormones triiodothyronine (T3) and thyroxine (T4) is frequently altered in rodent bioassays, which may be caused by the ability of chemical substances to induce xenobiotic metabolism in the liver. As a secondary consequence, the induction of xenobiotic metabolizing enzymes, such as uridine diphosphate glucoronyl-transferases (UDP-GT) or sulfotransferases, generally leads to an increase in the degradation of hormones. Mechanistic assays capable of analyzing this mechanism have not yet been standardized for endocrine disrupting chemicals, though work in this respect is underway and has been included in the OECD workplan for the Test Guidelines Programme. There is also still a lack of harmonized assays detecting disruption of hormonal transport, storage or release. Due to a number of research efforts e.g. by the Eurion cluster (https://eurion-cluster.eu/) in the EU, this deficiency will at least partially be overcome in the next ten years.

The way how to identify an ED has been described in several regulatory overviews such as the ECHA–EFSA guidance document for identification of endocrine disruptors⁷  or the OECD Guidance document 150.⁵  The latter document contains the so-called OECD Conceptual Framework for Testing and Assessment of Endocrine Disrupting Chemicals which is depicted in Figure 1.3 in a short version.

Generally, in regulatory ED assessment, the identification process starts with an examination of all data available for a specific substance, including data from regulatory studies as well as from the scientific literature obtained in a systematic review. Within this data package, evidence for adverse effects is detected and mechanistic information on potential endocrine modes of action is analysed and jointly considered for decision making. In the OECD Conceptual Framework (CF) these data are assigned to the first level (Figure 1.3).

While adverse effects can be detected in apical in vivo assays such as short-, medium- or long-term rodent bioassays according to OECD test guidelines (TG, level 4 and 5, Figure 1.3), mechanistic assays are required to obtain information on the mode of action of a substance. Mechanistic assays are frequently conducted in vitro to get information on whether or not a mode of action is endocrine (e.g. by interference with a hormone receptor or with hormone metabolism, level 2, Figure 1.3). Such assays may also be used for the screening of a large number of substances. There are, however, also two in vivo mechanistic assays that were published as OECD TG, namely the uterotrophic and the Hershberger assay that comprise level 3 of the OECD CF (Figure 1.3). While the uterotrophic assay is a short-term in vivo assay aiming at identifying (anti)estrogenic substances by analyzing female reproductive organs (OECD TG 440), the Hershberger assay is aiming at identifying (anti)androgenic substances by analyzing male reproductive organs after short-term exposure to respective substances (OECD TG 441).

All assays that have already been validated and harmonized at OECD level and that can give insights into endocrine effects, as well as a few other standardized tests, are included in the OECD CF. An evaluation of the endocrine aspects of these assays and proposals for further testing in case of positive level 2, 3 or 4 studies have been described in the OECD Guidance Document 150.⁵ 

While the higher tier studies such as short- or long-term studies in vivo are generally capable of analyzing adverse effects on all parts of the endocrine system, the existing mechanistic studies have a focus on the estrogen (E) and androgen (A) part of the HPG axis, on steroidogenesis (S) and on the HPT axis, abbreviated as thyroid (T). Hence, due to limitations in mechanistic assays, regulation of endocrine disruptors currently focusses on substances affecting the E, A, T and S parts of the endocrine system.

Short- or long-term animal studies conducted according to OECD testing protocols require the isolation and weighing of all organs as well as their pathological and histopathological evaluation. Table 1.1 shows an overview on parts of the endocrine system generally analyzed in these studies. In addition to the analysis of all endocrine organs, clinical chemistry analysis and hematology are performed. Analysis of some hormones, like the thyroid hormones T3, T4 and TSH have now been included in some study protocols like the 90-day feeding study in rodents (OECD TG 408) and the prenatal developmental toxicity study (OECD TG 414). Hence, by conducting in vivo studies a comprehensive overview of potential endocrine disruptive effects of a substance can be obtained.

In regulatory toxicology as well as in diagnosis of certain conditions, adverse effects are not considered in isolation but jointly. There are some effects that are considered diagnostic for endocrine disruption. Table 1.1 lists some of these effects, exemplified by data one would normally observe in a 90-day feeding study in rats, separated for E, A or S and T modalities. There are other effects considered to be indicative but not diagnostic of endocrine disruption. Examples of these effects consists of but are not limited to alterations in adrenal pathology or histopathology (e.g. hypertrophy of specific zones in the adrenal cortex responsible for biosynthesis of steroid hormones). Normally all of these effects would be considered together in a weight of evidence approach and several effects indicative of the same modality and adversity would give higher weight of evidence than isolated ones.

As previously mentioned, the in vitro assays included in the OECD CF give mechanistic information about the tested substances focused on E, A, T and S and are described in more detail in the respective paragraphs below.

A rather new concept in toxicology is the adverse outcome pathway (AOP) concept.⁸  It is a conceptual framework that presents existing information in such a way that linear relationships between triggering factors (molecular initiation events, MIE) and apical endpoints (adverse outcomes, AO) are created. It suggests that AOs at the level of the organism (for example tumors in hormone-responsive tissue or infertility) are caused by one (or more) MIE(s) (for example the activation of a hormone receptor, or the inhibition of an enzyme responsible for hormone metabolism) via several subsequent key events (KE) at the molecular, cellular and tissue level. Since to identify an ED not only have adverse effects to be detected in intact organisms but also the molecular mechanism has to be shown, this concept can help to bring these two aspects of endocrine disruption together.

Additionally, if MIEs are known, it is possible to screen for activity of substances against these MIEs. In the context of endocrine disruption such a MIE may be the activation or inhibition of the androgen receptor (AR) or of one of the estrogen receptors (ERα or ERβ). Multiple mechanistic in vitro assays exist to detect the potential of substances to activate or inhibit these receptors, some of which are already mentioned in Figure 1.3. They can be used to screen a large number of substances for their potential to interfere with the estrogen or the androgen part of the endocrine system. In the endocrine disruptor screening program of US EPA, for example, 52 pesticide chemicals (active and inert ingredients) were tested in the so-called Tier 1 Screening Battery that consists of five in vitro and six in vivo assays, all of which are also included in the OECD Conceptual Framework.⁵ 

A closely related but different program by US EPA is the Toxicity Forecaster (ToxCast) that uses assays that are not validated at the OECD level and therefore not listed in the OECD CF. In the ToxCast program, automated, robotics-assisted high throughput (HT) in vitro and in chemico assays have been developed and are used together with in silico tools for the characterization of substance properties including the detection of potential endocrine disruptive properties. So far, more than 1800 substances have been tested across more than 700 endpoints of data, including many that address endocrine properties. Additionally, the online database the CompTox Chemistry Dashboard offers data for more than 7000 other substances that have undergone some level of screening in the program.

Mechanistic in vitro assays demonstrating activity are one prerequisite, but not sufficient on their own to demonstrate the ability of a chemical substance to cause endocrine disruption. Another essential part is that the substance showing activity in vitro causes an adverse effect in vivo. In this context, it is essential that the substance reaches the endocrine target tissue in effect concentration levels. While AOPs were initially defined in a way that does not take toxicokinetics into account, toxicokinetic properties of a substances showing endocrine activity are very important to consider. These properties may be analyzed based on respective in vivo studies (e.g. OECD TG 417) or based on physiologically based pharmacokinetic (PBPK) modelling. Additionally, in vitro and in chemico assays that directly address different toxicokinetic aspects are in development and/or in validation.

In the following sections examples from the estrogen, androgen and thyroid parts of the endocrine system are presented along with schematic AOPs and available assays capable of screening for E, A or T activity.

In the past decades, it has become apparent that several endocrine-related diseases and disorders, many of which are connected to the estrogen pathway, are on the rise. For example, significant increases in the global rates of breast and endometrial cancer have been observed.⁹ 

The development of complex diseases like cancer depends, of course, on many external factors such as diet, nutrition status and smoking behavior in addition to genetic factors. It is therefore very difficult to estimate which additional unknown or unconsidered factors lead to the increase of disease rates. As per definition EDs disrupt endocrine pathways, thereby leading to adverse outcomes in intact organisms, it is important to identify potential EDs, because the exposure to these substances might contribute to the observed trends.

The schematic AOP pictured in Figure 1.4 lists different potential MIEs by which different (anti)estrogenic endocrine disruptive adverse outcomes (AO) may be caused. Estrogen receptor (ER) (ant)agonism, i.e. the binding of the substance to ER and associated activation (agonism) or blocking of endogen activation (antagonism) can function as MIEs, as well as all molecular events that alter the estradiol metabolism. Generally speaking, these MIEs lead, in turn, to altered expressions of estrogen-dependent genes and proteins, thereby altering cellular functions, leading to altered tissue functions. These alterations at the tissue level (e.g. hyperplasia in breast tissue as a consequence of estrogenic effects) can eventually result in adverse outcomes (e.g. breast cancer) at the organismal level.

This schematic AOP can be used to illustrate both the advantages and disadvantages of this way to represent information. On the one hand, the simplified linear fashion in which the MIE is connected via several KEs with an AO clarifies which assays targeting MIE or KEs should be used to systematically test if a substance might cause an AO via this specific route. On the other hand, however, this simple representation omits that the endocrine system is a complex system of feedback loops. Even though many chemicals may display certain (anti)estrogenic effects, in the in vivo situation these effects might be counteracted, thereby preventing an AO. The currently available assays, which are also listed in the OECD Conceptual Framework (Figure 1.3), focus mainly on the direct influence of the substances on the estrogen system, i.e. they address the MIEs outlined in Figure 1.4. It should, however, not be overlooked that the production and release of estradiol are regulated by the HPG axis and interference with this loop or crosstalk with other subsystems of the hormonal system may also lead to disturbances of the estrogen system. While several chemicals that are capable of affecting the estrogen system show direct interaction with the estrogen receptors ERα and ERβ (e.g. diethylstilbestrol, DES) or inhibition of aromatase (e.g. flusilazol) as suggested in the AOP there are other examples like atrazine that cause endocrine-mediated adversity, like mammary tumors, due to interference with the feedback loop. This is another reason that, despite all efforts to analyze estrogenic effects at the in vitro level, in vivo studies displaying the complexity of the whole organism are still an inevitable part of ED identification. Additionally, substances that interact directly with the ERs can have different affinities to ERα and ERβ, which might influence their (adverse) effects. This is further complicated by the fact that the effect of an ERα and/or ERβ interaction can be cofactor and thereby also tissue-dependent.

There are three in vitro and one in vivo OECD Test Guidelines (TG) that address different parts of the schematic AOP depicted in Figure 1.4. These TGs together with the endpoint they target are listed in Table 1.2.

OECD TG 493 encompasses two in vitro assays, in both of which the ability of a substance to bind recombinant (human) estrogen receptor alpha (hERα) is measured. This is accomplished by determining the capability of a substance to displace a radiolabeled ligand (17β-estradiol) from hER. Substances with possible agonistic or antagonistic ER properties can be detected with this assay, they can, however, not be distinguished. For this differentiation, the ER transactivation assays (ERTA, OECD TG 455) can be used. While the first assay of this TG, namely the Stably Transfected TA (STTA) assay only detects activation of hERα, the second assay, the VM7LucER TA assay, detects activation of hERα and hERβ together. An ERTA assay follows the principle of a reporter gene assay: If the ER (hERα or hERα/hERβ) is activated by a substance with ER agonist properties (or its activation is inhibited by a substance with ER antagonist properties) a reporter gene signal that can easily be detected is altered.

Beside ER (ant)agonism, alterations in the estradiol metabolism are also probable MIEs that may trigger AO (Figure 1.4). To test substances in this regard, the H295R steroidogenesis assay (OECD TG 456) that has been described above and that is capable of detecting substances affecting the biosynthesis of steroid hormone can be used.

Beside the OECD TG and the few other standardized assays listed in the OECD Conceptual Framework for Testing and Assessment of Endocrine Disrupting Chemicals, we have already briefly mentioned the assays that were developed in the US EPA ToxCast Program. For estrogenic activity, there are 18 in vitro high-throughput screening (HTS) assays in this program. It was acknowledged by US EPA that using the data from 16 HTS assays in combination with a suitable computational model for ER activity gives similar predictive results to the endocrine receptor binding, the estrogen receptor activation and the uterotrophic assays. Therefore, since June 2015, US EPA has accepted this ER model as a substitute for the aforementioned three assays in the US EPA EDSP Program (Tier 1).^10,11  Results described in recent publications indicate that as few as four HTS assays together with the appropriate interpretation model might be enough to deliver the same results as TG 493, TG 455 or TG 440.¹¹ 

Diethylstilbestrol (DES) has been very well studied in epidemiological studies and animal experiments. It is one of the first and most prominent examples of an ED.¹²  The potent estrogen was first synthesized in 1938. In addition to its use as a growth stimulus in cattle farming, it was administered to several million pregnant women in the USA, Europe, Asia and Australia from 1941 to the 1970s.¹³  Starting from the mid-1960s it became apparent that women who were exposed to DES in utero developed vaginal adenocarcinomas.

Further toxic effects of the prenatally administered DES on both men and women became apparent later – many of them with a very high prevalence. For example, abnormalities of the upper genital tract were diagnosed in about 69% of the women exposed prenatally.¹³  For all female reproductive organs, i.e. ovary, uterus and vagina, negative effects of in-utero exposure to DES have been described.

These effects could clearly be attributed to the estrogenic mechanism of DES, which is a potent activator of hERα and hERβ.

Some evidence exists that diseases related to androgen signaling are increasing.¹⁴  For example, in the past two decades an increase in testicular cancer has been observed as well as a significant increase in prostate cancer. While the causes of such diseases are multifactorial and lifestyle, smoking, overweight, nutrition or a lack of activity are major risk-factors, chemical substances may still play a role. Additionally, whether an increase in certain developmental conditions like cryptorchidism or hypospadias may be related to ED has been discussed (for an overview see IPCS 2012¹⁴ ).

The AOP pictured in Figure 1.5 shows potential MIEs that may lead to adverse effects with respect to androgen signaling. The (ant)agonism of the androgen receptor, as well as alterations in metabolism of testosterone can function as MIEs. These MIEs would as a consequence lead to altered expression of androgen responsive genes and proteins. This would consequently lead to altered functions of cells in androgen-responsive tissue. Depending on the nature of the effect, this may be proliferation or cell death, altered metabolism etc. Consequently, in the end, adverse effects at the organ or organism level would occur. Such effects could consist of tumors, malformations or reduced fertility.

While this schematic AOP illustrates how chemicals may display (anti)androgenic effects, it should not be overlooked, that the endocrine system is a complex system of feedback loops. Production and release of testosterone are regulated by the HPG axis and interference with this loop or crosstalk with other subsystems of the hormonal system may also lead to disturbances of the androgen system. Hence, while the assays presented below are capable of detecting the mechanisms described in the AOP, there may still be others that would not be detected.

A number of in vitro assays has been developed and is partially already validated and accepted at the OECD level. One such assay is the Stably Transfected Human Androgen Receptor Transcriptional Activation Assay (OEC D TG 458). This assay may be used for detection of AR agonist and antagonist activity of chemical substances. This assay follows the principle of a reporter gene assay. If the AR is activated by a chemical with AR agonistic properties (or its activation is inhibited by a substance with AR antagonistic properties) a reporter gene signal that can easily be detected is altered. Reporter gene assays, such as the one used in OECD TG 458, may also be used for screening. Another in vitro assay capable of screening of many substances is the H295R steroidogenesis assay (OECD TG 456) that has been described above and that is capable of detecting substances affecting the biosynthesis of steroid hormones. This assay may be used for the detection of substances affecting the biosynthesis of estradiol as well as testosterone (Table 1.3).

Linuron has been used as a herbicide. In a number of rodent bioassays generally required for the approval of a substance as a pesticide it has shown a number of adverse effects that are linked to ED (a brief summary is given in Marx-Stoelting et al. 2011¹⁵ ). Among these are many pathological and histopathological alterations in androgen-responsive tissues such as testis (reduced weight, hypoplasia, atrophy, Leydig cell tumors), adrenal gland (hypertrophy of the adrenal cortex), uterus (adenocarcinoma), ovaries (tumors), endometrium (tumors) that were observed in rats but partially also in mice and dogs. Developmental effects related to anti-androgenic properties were observed in respective developmental toxicity studies (e.g. Wolffian duct malformations). In a two-generation reproductive toxicity study the following adverse effects were observed that may be caused by an anti-androgenic mechanism: reduced fertility, infertility and reduced litter size. In subsequent in vitro studies, linuron was shown to have antagonistic effects on the androgen receptors of rodents and other species. Besides this, linuron has shown anti-androgenic properties in a Hershberger bioassay.¹⁶  Additionally, effects on biosynthesis of fetal testosterone and aromatase inhibition were observed in non-guideline studies. In summary, multiple adverse effects observed in several species and mechanistic data strongly indicated endocrine mechanisms (receptor antagonism and enzyme inhibition) as a plausible reason for the adverse effects of linuron. Thus, linuron can be considered as an endocrine disruptor.

Thyroid hormones thyroxine (T4) and triiodothyronine (T3) are involved in the regulation of fetal brain development in humans during the last two trimesters of pregnancy and late neonatal and postnatal brain development in rats. During early stages of pregnancy, a developing fetus completely relies on the maternal T4, becoming able to produce its own thyroid hormone from gestation week 11 in humans and during late gestation (GD18) in rats. Therefore, maternal hypothyroidism, which leads to low levels of circulating thyroid hormones, may increase the risk of delayed or abnormal cognitive development in offspring.¹⁷  Consequently, chemicals capable of disruption of thyroid function (TDCs) gained particular attention among regulators with a primary focus on reduction of potential exposure to TDCs during the sensitive stages of development. On the other hand, due to the rate-limiting role of iodide in hormone synthesis by the thyroid gland, adequate exposure to iodine via the diet became essential for prevention of mental impairment worldwide.¹⁸  A cheap and easy method of salt iodization was implemented worldwide to improve iodine intake, in spite of that 39 countries (out of 69 countries with data) reported iodine-deficiency in pregnant women in 2017.¹⁹ 

Other adverse endpoints associated with the hypothyroid state are increased risk of pregnancy loss and infertility in women. Additionally, in adults, thyroid hormones are involved in the control of metabolism, with hyperthyroidism (excessive function of thyroid gland) manifesting in metabolic disorders. The well-described risk factors for development of hyperthyroidism and thyroid cancers are a family history of Grave's disease and ionizing radiation, respectively.

Homeostasis of thyroid hormones is extensively regulated, starting from the hypothalamic–pituitary–thyroid axis, responsible for the synthesis and secretion of thyroid hormones into circulation followed by conversion into a biologically active form in target tissues (e.g. developing brain) and catabolism of hormones in liver. The regulation of the target genes of thyroid hormones on the cellular level occurs via nuclear thyroid receptors TRα1, TRβ1 and TRβ2. As of 2018, only a limited number of chemicals capable of binding to thyroid receptors had been discovered.²⁰  A simplified AOP network (see Figure 1.6) depicts a list of some MIE, key events observed at the circulation and tissue levels and the consequent AO, relevant for mammals.

Inhibition of enzymes involved in synthesis of thyroid hormones, thyroid peroxidase (TPO) and sodium iodide symporter (NIS) are well-accepted molecular targets. At the same time upregulated catabolism of thyroid hormones by UDP-GT via activation of xenobiotic receptors pregnane X receptor (PXR) and the constitutive androstane receptor (CAR) in liver still represents a topic in ongoing scientific debates. The debate stems from the discussion on human relevance of the liver tumours observed in the rat model, considering differences in xenobiotic receptor activation of the CAR and PXR between the general human population and rodents.²¹ 

Several OECD test guidelines include investigation of thyroid histopathology in mammals (OECD TG 407, 408, 414, 421/422, 443/416, 451–3) and currently more test guidelines include measurements of thyroid hormones in circulation than ever before (OECD TG 441, 407, 408, 414, 421/422, 443/416). While these may be sufficient for the identification of thyroid adversity, so far, the OECD framework lacks validated and harmonized in vitro assays suitable for investigation of the mechanisms of thyroid disruption. On the other hand, as of 2019, the Joint Research Center of the EU Commission (JRC) is coordinating a large-scale multi-center validation project to deliver a set of in vitro test methods that would cover different aspects of potential interference with the thyroid hormone system (See Table 1.4). Additionally, several published assays can be used for mechanistic investigations if the thyroid was identified as a target organ.

In the US EPA the task of the identification of thyroid disruptors is approached by implementing some new technological methods for the purpose of generating mechanistic information about environmental chemicals. Such technologies include, among others, HTS, frequently using robotic or semi-robotic technologies. To generate mechanistic data, HTS assays first need to be adapted from the original low-throughput format, to be tested on a smaller set of reference of test chemicals and only then can they be applied for the screening of large chemical libraries. Depending on the assay type, the screening might need to be carried out in a tiered approach with a first round of testing done at a single high-dose concentration followed by the testing at a range of concentrations and further being refined against the suitable cell viability assay.

The chemicals from ToxCast Phase I and II libraries are the best characterized mechanistically (see Table 1.5), which is of particularly value for the environmental pollutants that are frequently lacking the extensive toxicological characterization available for pesticides. Simultaneously, identification of the molecular targets of thyroid toxicity for some pesticides using HTS results contributes to the implementation of the novel EU legislation concerning endocrine disruptors.

PTU is a known goitrogen, i.e. a drug that inhibits thyroid hormone synthesis via inhibition of thyroid peroxidase and a currently approved drug for use in patients with hyperthyroidism.^30,31  In rodents, exposure to PTU during pregnancy leads to the maternal hypothyroidism as assessed by the levels of circulating T4 in pregnant dams and congenital abnormality in the periventricular region of the brain of progenies exposed during GD19-PND2.³²  Another known inhibitor of TPO, ethylenethiourea (ETU), is a transformation product of several dithiocarbamate pesticides. In the long-term toxicological studies, these pesticides induce thyroid cancer and in the reproductive studies with ETU, even low doses induced maternal hypothyroidism, reduced colloid content and displayed reproductive toxicity in treated dams.^33,34 

The way endocrine disruptors are regulated differs significantly between several fields of regulation and also in different countries. While in the EU hazard-based cut-off criteria have been introduced that do not allow active substances with ED properties onto the market for some classes of substances like pesticide active substances, no such cut-off criteria exist in the USA, Canada or major Asian countries. Additionally, while pesticides are regulated based on hazard identification only in the EU, the European Chemicals legislation does not provide for such a hazard-based cut-off. Under Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), endocrine disrupting properties may trigger identification of a chemical as a ‘substance of very high concern’ (SVHC). SVHC, which are included in the Annex XIV to REACH, are subject to authorization. Thus, such chemicals would only be allowed onto the market if they fulfil conditions for authorization, i.e. the risk from using the substance is adequately controlled. Unless it is possible to determine a threshold, it must be shown that the socio-economic benefits of using the substance overweight the risks and that there are no suitable alternative substances or technologies.

For other groups of substances, such as cosmetics or substances in toys, food contact materials or food additives, regulation is suggested to be updated soon. Since some substances are regulated by more than one regulation, it is important that the same principles apply across regulatory systems to stay in line with the concept ‘one substance – one toxicology – one assessment’.³ 

During the past decade, a lot of progress has been made towards the development of methods capable of detecting EDs as well as the development of criteria for risk assessment. This gives confidence that the level of protection from chemicals affecting the endocrine system has increased.

There is, however, some work to be done. The need for mechanistic analysis has forced the development of new in vitro methods. Along with the AOP concept, it seems reasonable to hope for a more mechanism-driven toxicity assessment in the future that makes use of more in vitro methods and less animal testing and is capable of screening of large numbers of substances.

The focus of endocrine disruptor research has previously been on substances interfering with reproduction (i.e. substances with estrogenic or androgenic properties or affecting steroidogenesis) and on substances affecting the thyroid. This has been abbreviated to EATS (estrogen, androgen, thyroid, steroidogenesis) and most assays developed so far and accepted by international regulatory bodies for hazard assessment of chemical substances focus on EATS-related mechanisms. Considering the frequency of endocrine related adverse effects in animal studies, it would also be justified to expand the focus to the hypothalamic–pituitary–adrenal (HPA) axis because some substances affect this axis.³⁵  Considering the frequency of endocrine disease in the clinical practice, it would also be justified to focus on substances affecting regulation of metabolism, potentially involved in the etiology of obesity or diabetes.

Accordingly, new methods for the detection of substances affecting non-EATS modalities driven by needs from clinical endocrinology are required.

Finally, harmonization of regulation of different classes of chemical substances is required and could be achieved by international cooperation in the future.