Toxins in Botanical Drugs and Plant-derived Food and Feed – from Science to Regulation: A Workshop Review

Abstract

In September 2022, the 3rd International Workshop on pyrrolizidine alkaloids (PAs) and related phytotoxins was held on-line, entitled ʼToxins in botanical drugs and plant-derived food and feed – from science to regulationʼ. The workshop focused on new findings about the occurrence, exposure, toxicity, and risk assessment of PAs. In addition, new scientific results related to the risk assessment of alkenylbenzenes, a distinct class of herbal constituents, were presented. The presence of PAs and alkenylbenzenes in plant-derived food, feed, and herbal medicines has raised health concerns with respect to their acute and chronic toxicity but mainly related to the genotoxic and carcinogenic properties of several congeners. The compounds are natural constituents of a variety of plant families and species widely used in medicinal, food, and feed products. Their individual occurrence, levels, and toxic properties, together with the broad range of congeners present in nature, represent a striking challenge to modern toxicology. This review tries to provide an overview of the current knowledge on these compounds and indicates needs and perspectives for future research.

Key words

Alkenylbenzenes - carcinogenicity - genotoxicity - pyrrolizidine alkaloids - relative toxic potency - risk assessment

Abbreviations

ALARA: as low as reasonably achievable

API: active pharmaceutical ingredient

AUC: area under the curve

BMC: benchmark concentration

BMD: benchmark dose

BMD: lower confidence limit

CI: confidence interval

CYP: cytochrome P450

DHP: dehydropyrrolizdine

DHPA: dehydro-PA

DSS: dextran sulphate sodium

EFSA: European Food Safety Authority

EMA: European Medicines Agency

GLP: good laboratory practice

γH2AX: histone 2AX, phosphorylated form

HMP: herbal medicinal product

HMPC: Herbal Medicinal Products Committee

ICH: International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use

iREP: interim relative potency as defined in 11

LC-MS: liquid chromatography–mass spectrometry

LC-MS/MS: liquid chromatography–tandem mass spectrometry

LogD: log distribution coefficient

NAM: new and alternative approach methodology

NOAEL: no-observed-adverse-effect level

NOEL: no-observed-effect level

OECD: Organisation for Economic Co-operation and Development

PANO: pyrrolizidine alkaloid N-oxide

PA: pyrrolizidine alkaloid

PBK: physiologically based kinetic

Ph. Eur.: European Pharmacopoeia

PHH: primary human hepatocytes

PoD: point of departure

REP: relative potency

rSCH: rat sandwich culture hepatocytes

SCCS: Scientific Committee on Consumer Safety

SULT: sulfotransferase
 

Introduction

In September 2022, the 3rd International Workshop on Pyrrolizidine Alkaloids (PAs) and Related Phytotoxins was held on-line, entitled ʼToxins in botanical drugs – from science to regulationʼ. As with the two preceding workshops, this workshop aimed at presenting new findings on the occurrence, exposure, toxicity, and risk assessment of PAs. In addition, it was intended to cover new scientific results related to the risk assessment of alkenylbenzenes, a class of natural plant constituents present in food, feed, and herbal medicines. This widening of the scope of the conference was in recognition of the increasing concerns and subsequent numerous scientific investigations related to natural alkenylbenzenes. The conclusions presented here attempt not only to summarize the reported findings across these two topics but also intend to highlight the presumptive consequences for future steps in risk management and regulation in the field of public health for both classes of compounds.

PAs are found in food of plant origin, food supplements, feed, and herbal medicines [1], [2], [3]. The reported contents may reach levels that raise a health concern and warrant a refinement of the current risk assessment. To this end, a better understanding not only of the mechanism of action of PAs but also of their structure-dependent toxicokinetic and toxicodynamic properties is key to such efforts. A refined species comparison (animal species vs. human) may help to explore the quantitative meaning of animal experiments for human risk assessment and improve the basis for a risk analysis for animal species including farm animals.

The study of cases of acute intoxication is of relevance for farm animals [4], but these cases may also reflect a risk for humans [1]. The possible risk from chronic exposure to PAs is a key question, e.g., for manufactured and/or traded food products, sometimes transported and distributed on a large scale. Here, the tracing of sources of contamination such as the origin of PAs present, e.g., in tea (Camellia sinensis L., Theacea) [5], one of the most important food items, needs special attention.

Both in humans and animals, the chronic toxicity of PAs is of particular interest since hepatotoxic and genotoxic events may accumulate over time and result in irreversible damage such as chronic liver failure or cancer (see below).

Asteraceae, Boraginaceae, and Fabaceae are the major plant families comprising large numbers of PA-synthesizing species [3]. Overall, more than 600 different PAs (congeners) have been found in nature, having as their major structural constituents a necine base (pyrrolizidine ring) and one or more necic acid(s) bound to the necine base via ester bridges. The PAs can be subdivided into monoesters, open-chained diesters, and cyclic diesters. All toxicologically relevant PAs bear a double bond in the 1,2-position of the necine base and require further metabolic oxidation of the pyrrolizidine ring to be activated to the reactive, toxic pyrrolic metabolites [6].

Damage to liver sinusoidal endothelial cells and hepatocytes is the most prominent adverse effect seen with PAs upon acute, sub-acute, and chronic exposure [7]. In humans and animals, necrosis of sinusoidal endothelial cells leads to veno-occlusive disease and liver failure [8], while pulmonary venous endothelial cells may also be affected. In laboratory rodents treated chronically with certain PAs, malignant tumors of the liver such as liver cell carcinoma and hemangio-endothelial sarcoma can develop [6]. Furthermore, the occurrence of extra-hepatic tumors, e.g., in the lung, the pancreas, and the intestine, has been described [1].

The reactive metabolites formed, i.e., the dehydro-pyrrolizidine (DHP) esters and/or the de-esterified DHP, the so-called pyrrolic metabolites, can undergo covalent binding to nucleophilic targets such as proteins and nucleic acids (reviewed in [7]). In fact, a number of PAs are genotoxic in bacteria, insects, and mammalian cells in vitro. Reaction with cellular targets is also considered responsible for non-neoplastic cell damage. The formation of reactive pyrrolic metabolites depends on an oxidative metabolism mostly catalyzed by members of the cytochrome P450 (CYP) enzyme family. Although hepatocytes are most probably the primary site of metabolic activation, secondary metabolites may be released from the hepatocytes and may affect other cell types. Furthermore, direct activation in other non-hepatic tissues may occur. The group of PA N-oxides (PANOs), representing products of an alternative but reversible type of oxygenation, has been the subject of recent investigations. Although not electrophilic themselves, PANO reduction can re-constitute the parent PA [9], which may then be activated following the classical ring-oxygenation pathway. Thus, the toxic potency of PANOs may be lower than that of the parent PA, although the difference is likely to depend on the actual conditions, such as the redox state of the chemical milieu, or on the reductase activities in intestinal bacteria and in tissues of the host.

Current risk assessment of PAs is focused on the formation of reactive intermediates being responsible for both cell damage and genotoxicity. The broad spectrum of PAs occurring in nature makes it difficult, however, to identify the relative toxic potencies of all individual PAs. Furthermore, a concise dose-response analysis of adverse effects including low, more relevant dose levels is challenging. The use of rodent data implies the need for a scientifically sound extrapolation to humans illustrating the requirements for an adequate use of human in vitro data in risk assessment. This includes the employment of toxicokinetic methodologies such as physiologically based kinetic (PBK) modeling. The application of in vitro data in the quantitative risk assessment of individual PA congeners, taking into account both their toxicokinetic and toxicodynamic properties, is a challenge being addressed but also a possible paradigm to apply to other groups of chemicals.

Currently, a carcinogenicity study in rats with the congener riddelliine is used as the basis for the assessment of the chronic cancer risk for all 1,2-unsaturated PAs 1, [10]. Since the marked differences in the cytotoxic and genotoxic properties between individual PAs have been widely described, this approach is scientifically problematic. Merz and Schrenk [11] have addressed this problem by suggesting interim relative potency (iREP) factors to describe the toxic and genotoxic potency of 1,2-unsaturated PAs. Briefly, this approach is based on fundamental structural considerations together with acute toxicity data from rodents, cytotoxicity data in mammalian cell culture, and genotoxicity data in Drosophila. The use of in vitro data on the toxicity and genotoxicity of PAs has been hampered, however, by a lack of data on their toxicokinetics in intact organisms. In addition, more work is needed on the detailed mode of action (MoA) of highly toxic and carcinogenic 1,2-unsaturated congeners. Furthermore, as mentioned previously, PANOs may need special considerations since the intestinal microflora are able to substantially convert N-oxides into the more toxic parent PAs after oral exposure. Additionally, N-oxides can also be reduced by the tissues and cells of the host [9].

Aspects on PA Occurrence: Stability and Transfer from Feed to Food Products

PAs are secondary metabolites produced by a wide variety of plants. They play an important role as defense compounds against attack by herbivores (insects and mammalians) and against fungal and bacterial pathogens. PAs occur in a wide structural variety, and it is estimated that approximately 2 – 3% of all flowering plants are capable of producing PAs.

As PA-containing plants have a global distribution, they will also be present in areas where feed commodities are grown. When not properly removed, these weeds will end up in the fodder for livestock or in ingredients used for animal feed. Not only can this lead to adverse health effects in livestock, it is also one of the routes by which PAs can enter the food chain [12]. After ingestion by the animal, PAs are exposed to specific conditions in the gastrointestinal tract. Metabolism will take place in the liver, followed by excretion via urine and feces. Nevertheless, traces of the PAs may eventually end up in products that are used for human consumption such as milk, eggs, and meat. Although the concentrations are expected to be rather low, due to the extensive metabolism in the animals, these products constitute an important part of the dietary intake of a large part of the population [13].

In the past decade, several studies have been conducted in which the transfer of PAs from feed to the animal-derived end products was addressed in more detail. Two studies looked into the transfer of PAs to cow milk [14], [15]. Both studies indicated limited overall transfer of PAs (less than 0.1%), but this transfer is strongly related to the chemical structure of the PA. [Table 1] presents an overview of the calculated transfer rates for individual PAs. Some PAs, such as jacoline and otosenine, show relatively high transfer rates, while for many other PAs, rates are very low, indicating extensive metabolism in the animal. PANOs were practically absent and some newly formed PA metabolites could also be tentatively identified. Very low transfer rates have been observed as well for tropane alkaloids, which are also esters [16]. In contrast, the transfer of quinolizidine alkaloids from lupin to milk is substantially higher, suggesting a higher resistance toward the metabolism of this alkaloid type [17]. PA transfer to eggs was studied by exposing laying hens to feeds mixed with small portions of five different plant species [18]. Transfer rates in the same range as observed for milk were obtained ([Table 1]). Jacoline and otosenine showed relatively high transfer rates, but this was also the case for retrorsine, usaramine, sceleratine, heliosupine, and rinderine. Overall, the transfer was low (< 0.1%), and almost no PANOs were detected in eggs.

Table 1 Pyrrolizidine alkaloid transfer rates from feed to milk and egg and uptake rates from feed by insect larvae, in % of administered or consumed amount.

Pyrrolizidine alkaloid	Cowʼs milk a [15]	Egg b [18]	Black soldier fly larvae c [19]	Lesser meal worm larvae c [19]
a Cows given by gavage ground material of Jacobaea vulgaris (common ragwort), Senecio vulgaris (common groundsel), or Echium vulgare (vipers bugloss); b Laying hens given pelleted feed containing J. vulgaris, S. vulgaris, S. inaequidens (narrow-leaved ragwort), Echium vulgare, or Heliotropium europaeum (common or European heliotrope); c Larvae raised on substrate coated with PA mixtures simulating the composition and concentration of the feeds used in (b)
Cyclic diesters
Erucifoline	0.004	0.055	2.1	1.5
Florosenine	0.30
Integerrimine	0.003 – 0.004	0.011 – 0.012	0.3 – 0.4	0.6 – 1.0
Jacobine	0.074	0.11	3.6	2.6
Jacoline	2.94	1.21	5.3	7.2 – 11.8
Jaconine	0.014
Otosenine	0.73	0.40
Riddelliine	0.017 – 0.083	0.074	1.2	0.6
Retrorsine	0.085 – 0.12	0.33 – 0.39	0.6 – 1.4	0.6 – 1.0
Sceleratine		0.98	1.8	3.9 – 5.3
Senkirkine	0.12	0.15	0.28	0.4
Senecionine	0.001	0.017 – 0.033	0.5 – 1.0	0.5 – 1.2
Seneciphylline	0.003 – 0.008	0.014 – 0.015	0.6 – 0.9	1.1 – 2.2
Senecivernine		0.042 – 0.043	0.4 – 0.5	0.7
Spartioidine		0.022
Usaramine	0.038	0.23 – 0.49	0.9 – 1.5	1.2 – 1.4
Open-chained diesters
Echimidine	0.29	0.013	0.7	0.5 – 0.8
Heliosupine		0.44
Lasiocarpine		0.002	0.6	1.2
Mono esters
Echinatine		0.037	3.2 – 4.4	28.5
Europine		0.15	4.1 – 4.5	50.8 – 63.0
Heliotrine		0.045	2.4	4.7 – 5.8
Lycopsamine	0.32
Rinderine		0.35	3.8 – 4.1	23.9 – 32.2

Very recently, the uptake of PAs by insects that are raised for food or feed applications have also been studied. Black soldier fly (BSF) larvae and lesser mealworm (LMW) larvae were raised on substrate spiked with simulated extracts of the five PA plants mentioned above [19]. It was noted that upon the addition of water to the feeds, the PA composition of the substrate changed; i.e., the PANOs were effectively converted to the corresponding parent PAs. The bioaccumulation factors were therefore based on the PA composition in the feed remaining at the end of the trial ([Table 1]). For LMW larvae, high transfer rates, in the order of 28 – 63%, were calculated for the heliotridine-type monoesters echinatine, europine, and rinderine. For the other PAs, as well as for BSF larvae, transfer rates were 10- to 100-fold lower.

In living plants, PAs are primarily present in their N-oxide form, as this facilitates their transport and distribution within plant tissues. However, physical, enzymatic, and microbiological factors may cause reduction or degradation of PANOs in biological matrices, e.g., during heating or storage [20]. The minimal transfer of PANOs to milk and eggs suggests a stability in the target animals that is even lower than that of PA-free bases. In vitro gastric rumen experiments showed that PANOs are efficiently reduced in the rumen to their parent PA form by microbial action within a few hours [15]. Subsequent metabolism of the parent PAs occurs in the liver. When excreted in milk, parent PAs are relatively stable, as thermal processing of milk and the preparation of dairy products had only a limited effect on the PA levels [21]. Microbial activity in the moist insect feeds was considered the most likely cause of why PANOs were efficiently reduced to the corresponding parent PAs.

In a general sense, these studies confirm the relatively low overall transfer of intact PAs from feed to the final products. Nevertheless, the studies also show that the transfer rates of individual PAs can differ strongly, and it thus matters which PA-containing weeds are present as contaminant in the feed.

Human CYP Enzymes Account for the Metabolic Activation of PAs

It has been stated previously that PAs require metabolic activation to exert their genotoxicity [22]. For retronecine- and heliotridine-type PAs, there are three major metabolic pathways: 1) the hydrolysis of the ester functional groups to form the necine bases and the necic acids; 2) the N-oxidation of the necine bases to form PANOs; and 3) the hydroxylation of the necine bases by CYP enzymes followed by spontaneous dehydration to produce dehydro-PAs (DHPAs), which are then hydrolyzed into 1-hydroxymethyl-7-hydroxy-6,7-dihydropyrrolizine (DHP) in biological systems ([Fig. 1]). The otonecine-type PAs requires an initial oxidative N-demethylation step, potentially catalyzed by CYPs, to produce DHPAs. Both DHPAs and DHP can react with DNA, generating adducts and inducing genotoxicity.

Fig. 1 Metabolic activation of lasiocarpine, riddelliine, and senkirkine by human CYP3A enzymes produces metabolites including DHP esters and DHP that react with DNA, leading to genotoxicity.

As for the specific types of CYPs, early animal studies showed CYP3A and 2B subfamilies were the major CYPs accounting for the biotransformation of PAs such as clivorine, lasiocarpine, and riddelliine [22]. By using human supersomes (microsomes prepared from insect cells infected with a virus engineered to express a CYP isoform), CYP3A4 and 3A5 were identified to be critical to activate retronecine-type PAs [23]. Monocrotaline was the only exception, which was mainly bioactivated by CYP2A6. Researchers have also been using mammalian cells transfected with CYP3A4 as a tool to evaluate the genotoxicity of various PAs [24], [25].

Some recent efforts have been undertaken to further identify the specific CYP enzymes that account for the bioactivation of PAs in genetically modified human lymphoblastoid TK6 cells [26]. By establishing a set of 14 TK6 cell lines, each expressing a single human CYP (CYP1A1, 1A2, 1B1, 2A6, 2B6, 2C8, 2C18, 2C9, 2C19, 2D6, 2E1, 3A4, 3A5, and 3A7) [27], the metabolism of lasiocarpine (heliotridine-type), riddelliine (retronecine-type), and senkirkine (otonecine-type) was evaluated. The LC-MS results revealed the formation of DHP, the main reactive metabolite of PAs, in CYP3A4-expressing TK6 cells exposed to all three PAs. DHP was also detected in CYP3A5- and 3A7-expressing cells after PA exposure, but to a much lesser extent. The other CYPs did not play a critical role in the biotransformation of the aforementioned PAs. The micronucleus assay, as well as the cell cycle analysis, also showed that PAs induced concentration-dependent genotoxicity in three CYP3A-variant-expressing TK6 cell lines. For relative potency analysis, a total of eight PAs were tested in CYP3A4-expressing TK6 cells ([Fig. 2]). All PAs that tested positive for micronucleus induction after CYP3A4 bioactivation were ranked according to their genotoxic potency, calculated by using Bayesian benchmark dose (BMD) modeling and a critical effect size of 50%. The results showed that lasiocarpine was the most potent inducer of micronuclei, with a benchmark concentration (BMC) of 0.02 µM, while lycopsamine was the weakest among the PAs tested, with a BMC of 736 µM.

Fig. 2 Bayesian benchmark dose (BMD) modeling evaluating the potency of PAs for micronucleus induction in CYP3A4-expressing TK6 cells. The BMD₅₀ estimates represent a critical effect size of 0.5 (50% increase above the controls). The bar represents the calculated lower and upper 95% confidence interval (CI) for each BMD value.

Similar to the results observed in TK6 cells, lasiocarpine was found to be the most potent PA in micronucleus induction in metabolically competent HepaRG cells, and lycopsamine was about two orders of magnitude less potent than lasiocarpine [28]. In HepG2 cells transfected with human CYP3A4, lasiocarpine and senkirkine were the most potent PAs in micronucleus induction, followed by seneciphylline, retrorsine, and riddelliine [25]. Heliotrine and lycopsamine, which are monoesters, were much weaker micronucleus inducers compared to the PAs with open or cyclic diesters in HepG2 cells. The potency ranking was almost identical (except for monocrotaline, which is metabolized by CYP2A6) to the findings in CYP3A4-expressing TK6 cells ([Fig. 2]), suggesting the genotoxicity results are highly reproduceable across different types of human cells.

Overall, concordant results using different human-based systems have demonstrated that CYP3A enzymes were the main CYPs metabolizing PAs and that the genotoxicity of PAs varies by several orders of magnitude. The field of genetic toxicology is at a crossroads, transitioning from the traditional “screen-and-bin” process to include data suitable for quantitative risk assessment. Comparing the potency of PAs is becoming an increasingly important component in risk assessment for regulatory decision-making, whereby the introduction of interim relative potency factors (iREP factors) can be a meaningful approach to effectively evaluate the risk of PAs based on their structures and the weight of genotoxicity evidence generated from different biological systems [11]. The details are described in the later sections (“Current status of the concept of relative potency factors for PAs: Perspectives and open questions”).

Effects of Pesticides on the Genotoxicity of PAs

Previously, the relative genotoxic potencies of 37 PAs were determined in vitro in the metabolically competent human HepaRG liver cell line using the γH2AX assay, and it was found that potencies differed by several orders of magnitude [29]. The γH2AX/HepaRG assay was also applied together with analytical methods (LC-MS/MS) to extracts of European heliotrope (Heliotropium europaeum), and this work demonstrated the usefulness of an effect-directed analysis approach to identify less-known but potent PAs [30]. More recently, the γH2AX/HepaRG assay was used to assess whether co-exposure to pesticides could modulate the genotoxic effects of PAs (Peijnenburg et al., in preparation). The rationale of the latter research question was that some pesticides are known to induce the expression of CYP3A4, which is one of the main enzymes catalyzing the conversion of PAs into their toxic metabolites, via which the genotoxicity of PAs is exerted. Furthermore, a study on a PA intoxication incident in Tigray, Ethiopia, demonstrated that the observed hepatotoxicity was likely to be caused by co-exposure to the PA 7-acetyllycopsamine and the CYP3A4-inducing insecticide DDT [31].

In the PA–pesticide interaction study, a set of pesticides was first screened for their ability to induce CYP3A4 by exposing differentiated HepaRG cells to the pesticides for 48 h, followed by measurement of CYP3A4 activity using a CYP3A4 assay (Promega). Several pesticides, such as fipronil, o,p’-DDT, p,p’-DDT, acetamiprid, and thiacloprid, were found to increase CYP3A4 activity. For example, treatment of the cells with 5 µM fipronil resulted in a 10-fold induction of CYP3A4 activity ([Fig. 3]). Subsequently, the effects of these pesticides on the genotoxicity of the PAs lasiocarpine, senkirkine, otosenine, and 7-acetyllycopsamine were determined with the γH2AX/HepaRG assay. In these experiments, HepaRG cells were first treated for 48 h without or with the pesticides, followed by a 24 h exposure to the PAs. Then γH2AX induction was determined using ≥ 1.5-fold above background as cutoff for genotoxicity, as applied previously [22]. Generally, all CYP3A4-inducing pesticides were shown to significantly increase the potencies of the studied PAs except for senkirkine in the γH2AX/HepaRG assay. This is illustrated for fipronil in [Fig. 4]. Altogether, the results suggest that certain pesticides (and other chemicals that induce CYP3A4) may potentiate the (geno)toxic effects of PAs upon-co-exposure.

Fig. 3 Induction of CYP3A activity upon exposure of HepaRG cells to increasing concentrations of fipronil.

Fig. 4 Effect of fipronil on the induction of γH2AX by PAs. HepaRG cells were exposed to AFB1 (aflatoxin B1; genotoxic control), Lc (lasiocarpine), Sk (senkirkine), AcLy (7-acetyllycopsamine), and Os (otosenine) without (left side) or with (right side) pretreatment with fipronil.

Whether such effects could take place at relevant human exposure levels depends on, amongst other factors, whether the critical internal concentrations required for CYP induction are reached, which requires further study.

Assessment of the Genotoxic Potency of PA N-oxides: Role of Oxygen Supply

Using the human hepatic cell line HepaRG, it has previously been demonstrated using the micronucleus assay [3] that the genotoxicity potency of 26 representative PAs varies significantly due to structural differences ([Fig. 5]). In addition, this work demonstrated that PANOs are less potent inducers of DNA damage than their corresponding parent PA when evaluated using standard in vitro experimental conditions with ambient air (21%) oxygen levels. However, one would expect PANOs to be more potent under low oxygen conditions that are representative of in vivo liver partial oxygen pressure since such conditions favor reduction reactions. In contrast, the parent PAs are expected to be less potent under these more biologically relevant low oxygen conditions when compared to ambient air, since their metabolic conversion to the DNA-reactive pyrrolic ester requires oxidative metabolism. To investigate the impact of oxygen status, heliotrine and its N-oxide were evaluated under low oxygen (2%) and ambient oxygen (21%) conditions in the HepaRG micronucleus assay using the experimental conditions described [28], and the formation of heliotrine from its N-oxide was measured analytically. For the low-oxygen-condition experiments, HepaRG cells were acclimated for one hour in a separate incubator maintained at 2% oxygen. Cells were then exposed to the compounds for 24 h under these low oxygen conditions. After treatment, cells were returned to ambient oxygen conditions and exposed to human epidermal growth factor for 72 h to promote cell division, during which micronuclei may form.

Fig. 5 HepaRG cells were treated with heliotrine (a) and its N-oxide (b), and the relative viability (lines) and the occurrence of micronuclei (bars) are shown. Grey lines/bars represent results from experiments performed under more physiological low (2%) oxygen (O₂) conditions, while black lines/bars represent results from ambient (21%) oxygen conditions.

Consistent with the expectation that metabolic activation to the DNA reactive form of the parent PAs will be reduced under low oxygen conditions, heliotrine showed a decrease in micronucleus induction under 2% oxygen conditions when compared to 21% ([Fig. 5 a]). The converse was true for heliotrine N-oxide, which showed an increase in potential for DNA damage at 2% oxygen compared to 21% oxygen ([Fig. 5 b]). These findings are in line with the expectation that low oxygen conditions favor reduction to the parent PAs and are further supported by the analytical results demonstrating that ~ 6× more heliotrine is formed from its N-oxide under 2% oxygen conditions compared to 21% oxygen (data not shown). While heliotrine and its N-oxide followed the expected pattern, initial data from further experiments seem to show that not all PA and PANO pairs follow the above outlined hypothesis regarding increased reduction under low oxygen conditions. Taken together, these initial data show that the use of physiologically more relevant lower oxygen levels for in vitro experiments tend to reduce the potency of PAs, while they tend to increase the potency for PANOs. Further work will be needed to better understand the possible impact of oxygen partial pressure on the PA risk assessment.

Relative Potency Factors of PA N-oxides at Realistic Low Dose Levels

PANOs are the predominant version of PAs in plants [32]. Upon consumption, PANOs are reduced mainly by intestinal microbiota and CYP enzymes in the liver to the parent PAs [9], which subsequently give rise to the formation of reactive pyrroles that can form protein- and DNA-pyrrole adducts. A REP_{PANO to PA} factor can be defined as the ratio between the potency of a PANO and that of the parent PA. Currently, PANOs are assumed to be either equally toxic to the parent PA (i.e., having an REP_{PANO to PA} factor of 1.0) [11], or to be substantially less potent than the parent PAs (REP_{PANO to PA} factors < 0.1), the latter mainly based on studies in in vitro models [22], [31], [33]. Thus, the toxicity and, thereby, also the REP_{PANO to PA} value of PANOs depend on the rate and extent of PANO reduction to the parent PA and other kinetic factors such as absorption that would influence what the liver is exposed to. Although the PANO reduction in the liver may be included in the in vitro cell model used, these models generally do not consider the PANO reduction by the intestinal microbiota, nor do they accommodate gastrointestinal uptake [34], [35]. Another factor that may influence the REP_{PANO to PA} values is that the in vivo studies from which these REP_{PANO to PA} values can be derived are generally performed at relatively high equimolar doses of the PANO and PA in order to enable detection of selected biomarkers, while estimated daily dietary intake for the human population occurs at dose levels that may be orders of magnitude lower [1].

The studies presented here were aimed at defining REP_{PANO to PA} values and the effect of dose levels on these values using new approach methodologies (NAMs) consisting of physiologically based kinetic (PBK) modeling and in vitro and in silico models to define the relevant PBK model parameters. The PBK model consists of a model for the PANO describing its kinetics, including the kinetics for PANO reduction by gut microbiota and liver CYP450s, and a sub-model for the kinetics of the corresponding PAs quantifying its general clearance and also 7-GS-DHP (a GSH conjugate formed from a reactive ʼpyrrolicʼ intermediate) formation. Based on the PBK model outcomes, various endpoints can be applied to define the REP_{PANO to PA} values. In a first method, the REP_{PANO to PA} values can be determined based on the ratio between the area under the curve (AUC) for the parent PA upon dosing either the PANO (plasma AUC_PA (dose PANO)) or an equimolar dose of the PA (plasma AUC_PA (dose PA ([Fig. 6], method 1) [36], [37]. In a second method, REP_{PANO to PA} values can be calculated based on the ratio of the amount of pyrrole-protein adducts upon dosing either PANO or an equimolar dose of PA ([Fig. 6], method 2) [38], [39]. The PBK model contained in silico-derived parameters and in vitro experimentally determined kinetic parameters and considered not only dose levels as high as the ones used in animal studies but also those that are low and more realistic for human exposure. The power of PBK modeling is the ability to predict REP_{PANO to PA} values at realistic low dose levels, which would be less accessible in in vivo studies, and also to predict data for humans by defining a PBK model for the PANOs in humans [37]. In vitro studies on 7-GS-DHP formation allowed definition of the kinetic parameters to include this metabolic step in the PBK model and enable calculation of the REP_{PANO to PA} value by method 2 [38], [39]. Riddelliine N-oxide, senecionine N-oxide, and their parent PAs were used as the model compounds so that predictions could be compared with available in vivo data [32], [40], [41], [42].

Fig. 6 Metabolic scheme of PANOs with senecionine N-oxide as an example and the two methods with different endpoints to calculate REP_{PANO to PA} values [38].

It was observed that the PBK-model-based predicted REP_{PANO to PA} values for riddelliine N-oxide [36] and senecionine N-oxide [37] in rats, calculated based on plasma AUC_PA levels (method 1), are close to in vivo-derived REP_{PANO to PA} values ([Fig. 7]) [36], [37]. Although the kinetic constants for PANO reduction and PA clearance in rats and humans were different, the two REP_{PANO to PA} values at low dose levels were found to be similar for both species [37]. Moreover, it appeared that the REP_{PANO to PA} values were dose- as well as endpoint-dependent at the oral dose ranges that were examined in this study (approximately 1 to 200 µmol · kg⁻¹ bw) ([Fig. 7 a] and [b]) [38], [39]. The decrease in the REP_{PANO to PA} value with increasing dose levels can be ascribed to the saturation of PA clearance upon high-dose PA and saturation of PANO reduction by intestinal microbiota upon high-dose PANO. Saturation of PA clearance increased the plasma AUC_PA upon high-dose PA, while saturation of PANO reduction reduced the plasma AUC_PA upon high-dose PANO. Together, these effects result in a decrease in the REP_{PANO to PA} value calculated as the ratio between the AUC_PA upon an equimolar dose of the PANO and the PA ([Fig. 6], Method 1). PBK modeling and in vitro studies on saturation of pyrrole glutathione conjugate formation provided insight into the potential mode of action underlying the lower REP_{PANO to PA} values obtained when calculated based on the amount of pyrrole–protein (method 2) than based on the plasma AUC_PA (method 1) [38], [39].

Fig. 7 Dose-dependent REP_{PANO to PA} value of (a) riddelliine N-oxide relative to riddelliine (10, 13) and (b) senecionine N-oxide relative to senecionine (11, 12) in rat, as obtained by PBK modeling prediction compared to study-derived values [32], [41], [42].

In conclusion, REP_{PANO to PA} values depend on kinetics (absorption, distribution, metabolism, and excretions (ADME)), dose (low versus high), species (rat versus human), and endpoint (formed PA versus pyrrole protein adducts). In the context of chemical risk assessment, ADME characterization provides critical information that can be applied to evaluate the potential hazards and risks associated with exposure and specific toxicological endpoints. With PBK-based modeling, REP_{PANO to PA} values at realistic low exposure levels could be defined and a species extrapolation from rat to human could be performed. It is important to note that to define REP values for the PANOs to be used in risk assessment for combined PA exposure, the obtained REP_{PANO to PA} values still have to be multiplied by the REP_{PA to RID} values (i.e., REP values of Pas compared to riddelliine as reference PA) to yield the final REP_{PANO to RID} values, ultimately resulting in REP_{PANO to RID} values lower than the REP_{PA to RID} values defined for the parent Pas. Ultimately, this work demonstrated the strength of using NAMs like PBK modeling, to replace, reduce, and refine (3R) the use of animal testing.

Support for Regulatory Assessment of Percutaneous Absorption of (retronecine-type) PAs through Human Skin

It has previously been demonstrated with PBK modeling how potency differences across a series of PA congeners can be informed for the oral route of exposure. There, in vitro measurements of metabolic fate and DNA adduct formation in rat sandwich culture hepatocytes (rSCH) were used alongside measurement of oral absorption and metabolic bio-activation relevant to the gastrointestinal tract [3]. Using this approach, it was confirmed that potency differences of PAs can span several orders of magnitude, as first proposed by Merz and Schrenk [11], with diesters demonstrating higher potency (1 – 0.1) than monoesters (0.01–>0.001). The respective PANO potency appeared to differ further still, by a factor of 2- to 10-fold, for diesters and monoesters, respectively, but challenges remain to better understand the fraction that reaches the liver due to complexities of a conversion to the free base by gut microbiota [3].

More recently, attention has turned to considering the dermal route of exposure and the extent to which PAs could be absorbed and/or metabolized in the skin. This is necessary for the risk assessment of some topically applied herbal medicines, namely those that naturally contain PAs and where, in spite of additional manufacturing steps to minimize levels, a low-level presence can be expected to remain. However, when individual PAs were taken into consideration for the risk assessment, it became apparent that measured in vitro data to understand intrinsic potency did not exist for all compounds of interest. Formerly, a ratio of DHP-DNA adducts/in vitro AUC was used, which was experimentally determined using the in vitro rSCH model to provide a measure of hepatocyte exposure to reactive metabolites and thereby inform intrinsic potency [43]. These experiments require complex in-house synthesis of DHP-DNA adducts for use as authentic analytical standards for DNA adduct quantification, and measurements can be very challenging, particularly for weakly potent PAs. Given the importance of lipophilicity and ionization state on the kinetic processes of a chemicalʼs absorption and disposition [44], it has been suggested that the degree of ionization and lipophilicity may reflect potency differences between various PAs and respective PANO forms. Further investigation of a series of PAs compared both predicted and experimental LogD (decadic logarithm of the distribution coefficient between octanol and water) values with previously measured DNA adduct/AUC values ([Fig. 8], manuscript in preparation). The predicted LogD values compared well to the experimental LogD values (R² > 0.9), which in turn did show a strong relationship to DNA adduct/AUC measurements, in line with the linear relationship that was determined previously ([Fig. 9]). This relationship can be explained by marked differences in physical–chemical properties between these compounds, which are known to influence chemical permeation mechanisms (ionization) and metabolism (lipophilicity), whereby hydrophilic chemicals with the highest basicity demonstrate minimal metabolism and relatively low DNA adduct formation and, in contrast, lipophilic weak bases, which are more likely to permeate through the cell membrane and exhibit higher metabolic loss and DNA adduct formation. The extent of ionization and lipophilicity was therefore confirmed to have an overall influence on hepatocellular disposition and metabolic activation potential.

Fig. 8 Regression model correlation between DHP-DNA adduct/in vitro AUC from rSCH model and experimentally measured Log D values. MCT = monocrotaline.

Fig. 9 Regression model correlation between DHP-DNA adduct/in vitro AUC from rSCH model and Log D predictions from ACD/Percepta software.

On the basis of this relationship, a regression model was generated to correlate DHP-DNA adduct/in vitro AUC values from the in vitro rSCH model and Log D, allowing for the use of LogD values to predict DNA adduct/AUC values for ʼuntestedʼ compounds and inform as a surrogate for intrinsic potency. When scaled relative to riddelliine, the reference PA used in establishing regulatory limit values, both the measured and predicted intrinsic potency values for individual Pas, was applied in a risk assessment example for a topical herbal medicinal product. Based on an assumption of 100% bioavailability via both the oral and dermal routes of exposure, the resulting margin of exposure was calculated to be considerably greater than the minimally accepted 10 000-fold. Notwithstanding this critical observation, it was hypothesized that the dermal route of exposure would also serve to limit bioavailability, and so, work was undertaken to assess the penetration of PAs into and through human skin using a GLP and OECD test guideline-compliant protocol [45]. The dermal absorption of the compounds investigated, representative of retronecine-type monoesters (lycopsamine) and open-chained diesters (echimidine), as well as cyclic diesters (retrorsine), revealed a low level of penetration through human skin. Under finite exposure conditions, recovery of each PA at 24 h post-exposure was greatest in stratum corneum > epidermis > receptor fluid > dermis. In the absence of regulatory guidance for the assessment of dermal penetration of herbal medicinal products, the 2017 EFSA guidance for pesticides was used for the quantitative interpretation of the data [46]. Although differences to other guidances exist, e.g., from the OECD [47], [48] and SCCS [49], dermal absorption values were derived corresponding to the sum of the receptor fluid, receptor chamber wash, skin (epidermis + clingfilm + dermis), and tape strips (3 – 20) of the stratum corneum. To address the variability between replicates, the EFSA guidance was followed using a multiple of the standard deviation added to the mean dermal absorption value, determined by the number of replicates used (n = 11). Even with this highly conservative approach to determine skin absorption, low dermal penetration values of < 10%, across all the PAs tested, were derived.

Initial experiments were also conducted to assess metabolic fate in human skin across a series of PAs. These investigations were conducted over a 4 h time period using human skin S9 at a protein concentration of 1.0 mg protein/mL and a substrate concentration of 10 µM, with typical co-factors to aid phase-I-dependent catalytic activity. No apparent losses were detected for the monoesters lycopsamine or intermedine or for the open diesters 7-acetylintermedine, 7-acetyllycopsamine, and echimidine.

Given the observed low penetration and low first-pass effects in skin, further adjustments in the risk assessment of dermal exposure to PAs may be possible. This will require refinement of the reference value used in the risk assessment, in this case the BMDL₁₀ for riddelliine. In the case that oral bioavailability of riddelliine is less than 100%, then refinement to the reference value used for risk assessment would be needed to complement the use of human dermal exposure adjusted for bioavailability. Further work is needed to explore this, but in the meantime, an understanding of intrinsic potency differences, measured or modeled using LogD values, may assist in the risk assessment of PAs across different routes of exposure.

Control of PA Impurities in Herbal Medicinal Products – Perspective of a National Regulatory Authority

Exposure to naturally occurring PAs has been associated with a risk for adverse health effects in humans and animals, due to their known hepatotoxic and potentially carcinogenic effects. In addition to plants, which produce PAs themselves, presumably as a defense mechanism against insect herbivores, the contamination of plants with PA-containing weeds has been identified as a source for PA uptake [50].

The presence of PAs in plants and derived herbal medicinal products (HMPs) requires action of the pharmaceutical companies and regulatory authorities. From a regulatory perspective, graduated plans and public statements by regulatory authorities and their associated committees have been primarily used to address necessary restrictions and requirements. Guidelines and requirements have been continuously adapted to the prevailing state of scientific knowledge, especially, but not exclusively, concerning contamination of plants without natural PA expression [50], [51], [52].

The requirements for HMPs comprise, for example, PA intake limits in release specifications with respect to the daily dosage and age group (e.g., 1.0 µg total PAs/day for adults). In addition, specific requirements for analytical methods and their validation are listed. Guidance documents and monographs have been published to illustrate the approaches and requirements for the challenging trace analysis of PA contaminations [52], [53], [54]. In this context, the growing knowledge and experience of different stakeholders can also be considered to modify requirements on the basis of the collected data. Based on the growing knowledge and evaluation of carcinogenicity data, recommendations on the maximum daily intake limits for PAs have been published and adjusted over time [50]. With the end of the previous transitional arrangements concerning the recommended maximum daily PA intake, the required introduction of age- and weight-based PA specification limits for HMPs with pediatric indication can currently be seen as an important challenge [50]. Appropriate adjustments to affected HMPs will, in many cases, require further actions by pharmaceutical companies and regulatory authorities [50]. In conclusion, the publication of more and more relevant PA-specific guidance documents, as well as the increased experience and the ongoing engagement of agricultural producers of medicinal plants, pharmaceutical companies, and test laboratories in the implementation of relevant measures, continues to improve the control of PA contaminations.

The Key Role of the Gut–liver Axis in PA-induced Hepatotoxicity and Enterotoxicity

Outbreaks of acute PA poisoning cases have been reported worldwide due to the ingestion of PA-containing herbal remedies [55] or PA-contaminated food products [56], [57], but the intoxication outcome of long-term exposure to relatively low levels of PAs remains largely unknown. Dehydro-PAs (DHPAs), reactive PA metabolites, bind to proteins and form protein adducts, thereby leading to cytotoxicity [58], [59]. Apart from causing tissue damage in the liver where PAs are primarily metabolized, the extra-hepatic toxicity of PAs has been less investigated.

The gut–liver axis not only refers to the anatomic connection between liver and gut but also to their close coordination in various biological processes. The role of the gut–liver axis in PA intoxication and the underlying mechanism was studied using mice orally treated with retrorsine, a representative toxic PA, at 20 mg/kg bw per day for 14 weeks. Such long-term PA exposure was found to induce intestinal injury manifested by intestinal epithelium damage and disruption of intestinal barrier function. Furthermore, using mice with tissue-selective ablation of CYP activity, it was found that hepatic CYPs, but not intestinal CYPs, predominately catalyzed the PA bio-activation to generate reactive DHPAs, which are further transported via bile excretion into the intestines to exert enterotoxicity ([Fig. 10]). Additionally, the impact of gut-derived pathogenic factors in retrorsine-induced hepatotoxicity was investigated in mice with dextran sulphate sodium (DSS)-induced chronic colitis. Compared to retrorsine-exposed normal mice, DSS-induced colitis mice demonstrated exacerbated retrorsine-induced liver injury manifested by enhanced hepatic vasculature damage, fibrosis, and steatosis. Furthermore, DSS-induced colitis increased the gut permeability; therefore, more gut-derived endotoxins were transported into the liver. On the other hand, DSS treatment did not alter hepatic CYP activity but significantly reduced the hepatic glutathione level, thereby suppressing the PA detoxification pathway.

Fig. 10 Schematic illustration of PA intoxication involving the gut–liver axis. The orally ingested PAs are initially absorbed in the intestines and transported into the liver, where they are metabolized into toxic metabolites, DHPAs, which cause hepatotoxicity. In addition, the hepatic-derived DHPAs are also transported via the bile duct into the intestine, thereby leading to enterotoxicity.

Taken together, these findings provide the first evidence of PA-induced intestinal injury as a consequence of liver metabolism and highlight the importance of gut homeostasis in PA-induced hepatotoxicity [60]. These findings warrant public awareness and further investigations of PA-induced hepatotoxicity associated with chronic intestinal disorders.

Potency Ranking of PAs Using a Genotoxicity Test Battery in Human Liver Cells

The genotoxic action of 1,2-unsaturated PAs comprises DNA adducts, strand breaks, and crosslinks [6], [58], [61], [62]. Previous studies showed that the chemical structure of PAs determines their reactivity and hepatotoxicity (summarized in [3]). However, there is a lack of quantitative genotoxicity data, particularly in primary human hepatocytes as relevant target cells. The objective of this study was to analyze the impact of the chemical structure on the genotoxicity and cytotoxicity of PAs and to derive the genotoxic and cytotoxic potency of PAs in different human liver cell models.

A panel of 11 PAs comprising monoesters (heliotrine, lycopsamine, europine, and indicine), open-chained diesters (echimidine and lasiocarpine), and cyclic diesters (riddelliine, senecionine, seneciphylline, retrorsine, and monocrotaline) was selected for our study. To this end, genetically engineered human HepG2 liver cells with CYP3A4 expression [63] and primary human hepatocytes (PHH) were used. As toxicological endpoints, cell viability/cytotoxicity and different genotoxicity markers including DNA strand-break induction (comet assay), accumulation of the p53 tumor suppressor protein, and formation of γH2AX were analyzed. The results showed a clear structure-dependent cytotoxicity for PAs in HepG2-CYP3A4 cells and allowed for cytotoxic potency ranking, which was corroborated in PHH ([Fig. 11]). Furthermore, our genotoxicity test battery consistently revealed the structure-dependent genotoxicity of PAs in HepG2-CYP3A4 cells and allowed for genotoxic potency ranking via BMD modeling, which correlated closely with the determined cytotoxic potency. Importantly, these findings were confirmed in PHH and further provided evidence for genotoxic effects even in the absence of cytotoxicity ([Fig. 12]) [64]. In summary, the work in human liver cell models, including the gold standard PHH, strongly supports the concept of grouping PAs according to their toxicity as first proposed by Merz and Schrenk [11], substantiating the provisional interim relative potency factors [3].

Fig. 11 EC₅₀ values determined in HepG2-CYP3A4 cells and primary human hepatocytes (PHH) after 24 h incubation with PA monoesters (ME), open-chained diesters (DE), and cyclic diesters (CE) based on cell viability measurements. *indicates that EC₅₀ couldnʼt be determined due to weak cytotoxicity. Please note that 11 PAs were studied in HepG2-CYP3A4 cells, while 6 PAs were tested in PHH. Data were taken from [59].

Fig. 12 Genotoxic potency ranking of PAs in HepG2-CYP3A4 cells and primary human hepatocytes (PHH) using the endpoint γH2AX. Concentration-response data were used to derive BMD values with 90% confidence intervals (CI) by PROAST. PAs were ranked according to their genotoxic potency based on the derived BMD values. Depicted are the BMD confidence interval plots in HepG2-CYP3A4 cells (black) and PHH (green). Please note, 11 PAs were studied in HepG2-CYP3A4 cells, while 6 PAs were tested in PHH. Data were compiled from [64].

Current Status of the Concept of Relative Potency Factors for PAs: Perspectives and Open Questions

An increasing number of publications provide convincing evidence that the toxicity of PAs differs considerably between congeners. Attempts to take these differences into account led to the derivation of interim relative potency factors (iREP factors) for 15 PA congeners and 3 PANOs [11]. This provisional assignment of factors to PAs was based on their combined genotoxic potency in Drosophila, cytotoxic potency in vitro, and acute toxicity in adult rodents. This approach also tried to identify certain structural features associated with the iREP factors in order to allow a future classification of congeners with limited or no experimental data, including the differences in potency between and within the major structural classes of monoesters, cyclic diesters, and open-chained diesters.

In general, chronic liver damage and liver tumor formation have to be considered as particularly critical endpoints [3]. Furthermore, the assessment should consider both acute toxicity due to accidental intake of PA-plant material and chronic exposure to much lower doses eventually leading to these endpoints in a time- and dose-dependent manner. Currently, large sub-chronic or chronic animal experiments with a relevant number of congeners to explore their dose-response relationships are not feasible. Thus, there is a need for the application and refinement of relevant NAMs as the major source of additional information. These should comprise models for cytotoxicity, genotoxicity, and toxicokinetics. Data from such studies can be used for the modeling of internal exposure and the resulting risk, as well as structure-potency and mode-of-action studies. However, achieving a consensus among responsible regulatory bodies in order to better define the requirements and conditions for the regulatory use of such data needs to be considered.

From a scientific point of view, investigations into the reasons for the pronounced differences in toxicity between congeners are of utmost importance. Recent in vitro metabolism studies in human and rat liver microsomes [65], [66] and in rat hepatocytes and CYP3A4-expressing human HepG2 cells with a broad spectrum of PA congeners revealed that their relative toxic potency is related to the extent and rate of metabolism of the parent PA [67]. Thus, highly potent congeners such as lasiocarpine or riddelliine are extensively metabolized, whereas several monoesters with low toxic potency are particularly resistant to metabolic degradation ([Fig. 13]). These findings were supplemented by detailed analyses of the metabolic patterns showing that those extensively metabolized, highly toxic PAs also form detectable amounts of one or more GSH adducts, while the monoesters did not [67]. Taken together, these studies indicate that the wide differences in relative potency between PAs are structure-dependent and are due to pronounced differences in both the extent and type of the metabolic conversion of the parent PA. Finally, the fraction of reactive metabolites available at the targets is the major driver of damage and thus of toxicity. This conclusion also implies that differences in cellular uptake between congeners seem to play a limited role, since the aforementioned, striking differences were also seen in microsomes, i.e., independent of cellular uptake.

Fig. 13 Remaining PA concentration (%) after incubation with human liver microsomes and glutathione for 360 min. Taken from [66], where experimental details are provided.

With these data, it appears warranted to take iREP factors into account in the risk assessment of PAs. This is also true for monocrotaline, which exerted a much lower toxic potency in several in vitro assays than its cyclic diester structure would suggest [11]. Metabolism data, however, provide evidence that monocrotaline is metabolized as slowly as other PAs with low potency [65]. Although exhibiting a cyclic diester structure, the diester ring is distinctly different from all other cyclic diesters ([Fig. 14]), probably explaining these findings. These data suggest that an adaptation of the iREP factor for monocrotaline (and related structures) to, e.g., a value of 0.1, may be warranted.

Fig. 14 Chemical structures of two cyclic diester PAs, monocrotaline (11-membered ring) versus riddelliine (12-membered ring).

Using iREP factors, instead of summing up all PAs under the assumption of equal potencies, for the calculation of total PA levels, human exposure, etc., was reported to lower the calculated levels by 2- to 3-fold [68]. However, there are cases where PAs with low potency are by far the major contributors to the total PA content of certain plants, strongly enhancing the impact of using individual iREP factors. A good example is lycopsamine and related PAs in borage, where the use of iREPs may reduce the calculated risk estimates and, hence, would increase the permitted intake by more than 100-fold [69]. Thus, it is recommended to consider the use of iREP factors for PAs in medicinal and food preparations to avoid possibly inadequate regulatory restrictions for the use of food, feed, and medicinal products contaminated with a large contribution of PAs with low toxicity to the overall PA levels.

Novel Approaches in the Risk Assessment of Alkenylbenzenes

Alkenylbenzenes are secondary plant metabolites that occur in various herbs and spices, like basil, fennel, and parsley. Exposure to alkenylbenzenes is a result of a direct consumption of these herbs and spices as food, the use of essential oils as flavoring ingredients, or the use of herbal food supplements with ingredients containing these compounds. In addition, there is an increasing interest in the use of alkenylbenzene-containing essential oils as feed additives, which can potentially lead to transfer from feed to food. Given the genotoxic and carcinogenic effects of various alkenylbenzenes, the presence of these substances in food and feed poses a health concern.

Several challenges exist with the risk assessment of alkenylbenzenes. Carcinogenicity studies are, for example, performed at high dose levels in rodents, and questions can be raised with respect to how these results are best extrapolated to low-dose human exposure scenarios. In addition, human exposure to the pure chemical does not occur, but it occurs in a complex matrix of substances. Furthermore, while estragole, safrole, or methyleugenol are known to be genotoxic and carcinogenic based on animal experiments, less information is available for other alkenylbenzenes like myristicin or elemicin.

NAMs play an important role in filling up data gaps in the risk assessment of alkenylbenzenes ([Fig. 15]). Given that the genotoxicity and carcinogenicity depend on the metabolic activation of these substances, particularly NAM-derived information on metabolism can be used to investigate the relevance of the observed effect(s) across species and to inform and/or refine the risk assessment of alkenylbenzenes.

Fig. 15 NAMs to study species difference, human variation, and matrix modulation of bio-activation and detoxification of different alkenylbenzenes.

[Fig. 15] shows the NAM that has been set up at Wageningen University to study species differences, human variation, and matrix modulation of the bio-activation and detoxification of different alkenylbenzenes. In vitro measurements are performed to derive the kinetic constants for the different bio-activation and detoxification pathways. Upon integration of these kinetic data in a physiologically based kinetic (PBK) model, one can simulate the bioactivation and detoxification in an animal species of interest. With this approach, it could, for example, be determined that dose-dependent effects and species differences in bio-kinetics cannot be used as arguments to reduce the default uncertainty factors in the risk evaluation of alkenylbenzenes [70], [71], [72], [73]. Also, differences in the extent of bio-activation between different alkenylbenzenes could be established [73]. Furthermore, the flavonoid nevadensin was identified to inhibit sulfotransferase (SULT)-mediated bio-activation and DNA adduct formation. However, this matrix-based interaction appears to be dose-dependent and predicted by the PBK models to be absent at realistic low dietary human intake [74], [75]. Overall, in vitro kinetic experiments in combination with PBK modeling provide a relevant framework to explore the dose-dependent effects, species differences, human variation, and matrix modulation of alkenylbenzenes.

Alkenylbenzenes and Medicinal Products

Alkenylbenzenes occur in a large number of plant families with relevance for medicinal products, especially in Apiaceae, Asteraceae, Rosaceae, Solanaceae, Piperaceae, Rutaceae, or Poaceae. Because of the volatility of the majority of alkenylbenzenes, they are found in the respective essential oils, although they may also occur as nonvolatile [76], [77]. The scientific interest in the toxicology of alkenylbenzenes was stimulated mainly by the discovery of toxic properties of safrole in the 1950s and 1960s [78]. Because of their favorable sensory properties, pure alkenylbenzenes or the essential oils containing them have been used in industrial food production for decades. The increasing knowledge of their toxicological profiles prompted regulators in the USA to prohibit the use of safrole already in 1960, of beta-asarone in 1968, and, only recently, of methyleugenol in 2018. In the EU, the use of pure safrole, asarone, methyleugenol, and estragole in food was prohibited by regulation (EC) 1334/2008, effective since January 2011 [79]. With this same regulation, permissible concentrations in certain food categories are defined for these alkenylbenzenes if naturally present in flavorings and food ingredients with flavoring properties and in certain compound foods as consumed, to which flavorings and/or food ingredients with flavoring properties have been added. This regulatory action was preceded by safety assessments of the respective compounds by the Scientific Committee on Food [80] and the Council of Europe [81]. The Herbal Medicinal Products Committee (HMPC) elaborated statements on the use of methyleugenol and estragole in HMPs [82], [83], [84], [85]. Between 2005 and 2013, numerous publications on estragole metabolism, toxicity, and mode of action, as well as estragole occurrence levels, particularly for fennel teas, appeared in the scientific literature, and the European ban of four pure alkenylbenzenes in food was published in 2008. In 2011, an NTP study on sub-chronic toxicity and carcinogenic effects (of note: no full carcinogenicity study) of estragole was published [86]. This prompted the HMPC to start an assessment in September 2013. The draft revision document as of November 2014 proposed an acceptable daily dose of 0.5 mg/person per day (adult, 50 kg bw), derived with reference to a BMDL₁₀ of 10 mg/kg bw/d, as previously discussed by EFSA and considering that a NOAEL would be lower, by application of an additional safety factor of ten [87]. In the further course of the revision process, the HMPC considered newly published regulatory guidance relevant to the issue [88]. In addition, the awareness increased that estragole-containing preparations are not only relevant as active pharmaceutical ingredients (APIs) of HMPs but also as excipients of many HMPs, as well as of numerous chemically defined medicinal products, necessitating further alignment with the European Medicines Agency (EMA) standard procedures. The second draft (Feb. 2020) no longer defined a limit dose for the general population but recommended the use of the ALARA principle, except for children under 12 y and pregnant/breastfeeding women. Furthermore, the principles of ICH M7 (less-than-lifetime principle, food background exposure, and the TD₅₀ approach in the absence of a robust BMD basis) were fully applied, and estragole-containing excipients were also considered. The final public statement corroborated these approaches and was published on 1 March 2022 [89], [90].

On 20 July 2022, the HMPC has drafted revisions of the monographs on bitter fennel fruit and sweet fennel fruit, as well as of bitter fennel oil. The HMPC no longer endorsed the use of bitter fennel oil in HMPs due to the high dosage specified in the hitherto existing HMPC monograph and the resulting high estragole intake. As regards the fennel fruit monographs, the HMPC proposed a reduction in the single/daily doses to the minimum levels indicated in the previous versions, reflecting the ALARA approach promoted in the public statement on estragole [91], [92], [93].

While limited to a max. 5% of fennel essential oil by the respective Ph. Eur. monograph, estragole is typically present in the essential oil portion of bitter fennel fruit at levels > 2.5%. The HMPC public statement on estragole suggests selection of “low estragole cultivars” as an option for minimizing the estragole content of fennel fruit. However, despite extensive breeding and selection efforts, it was not possible until now to establish fennel cultivars with estragole levels significantly lower than 2.5% (with reference to the essential oil portion) without serious impairment of vitally important properties (e.g., pest resistance toward Mycosphaerella and frost resistance). In addition, a strong positive correlation was observed in these studies between the content in estragole (Ph.Eur.: max. 5%) and anethole (Ph.Eur.: min 60% content in the essential oil). Once the estragole content undercut a level of 2.0 (2.2%, respectively), no single plant yielded fruit compliant with the pharmacopoeial requirement for anethole [94].

Although the HMPC public statement on estragole and the respective monograph revision drafts continue to principally allow the use of estragole-containing HMPs and fennel fruit in children, it will be extremely difficult if not impossible for many existing products to meet the guidance value.

Owing to the widespread occurrence of estragole in food plants (for examples, see [Table 2]), estragole background exposure can be considered significant.

Table 2 Occurrence of selected alkenylbenzenes in plants used as medicine and/or food.

Plant	ES	ME	SAF	ASA	MYR	HMP	Food
ES = estragole; ME = methyleugenol; SAF = safrole; ASA = β-asarone; MYR = myristicin; HMP = herbal medicinal products
Acorus calamus				x		x
Anthriscus cerefolium	x						x
Artemisia dracunculus	x	x				x	x
Boswellia serrata	x					x
Citrus aurantium	x						x
Cymbopogon sp		x
Foeniculum azoricum	x						x
Foeniculum vulgare	x	x				x	x
Fragaria vesca	x						x
Glycyrrhiza sp.	x					x	x
Hippophae rhamnoides
Illicium verum	x	x	x		x	x	x
Malus domestica	x						x
Mangifera indica	x						x
Musa sapientium	x						x
Myristica fragrans		x	x		x		x
Ocimum basilicum	x	x			x		x
Pastinaca sativa					x
Petroselinum crispum			x				x
Pimenta dioica	x						x
Pimpinella anisum	x	x				x	x
Piper nigrum	x	x	x		x		x

However, robust quantitative data on background exposure to estragole via food is not available today. The EFSAʼs assessment, as recently mandated by the European Commission, will hopefully improve the information basis about estragole exposure from food intake [95]. The EFSA must provide the output of its assessment by 15 May 2025. For the coming years, the challenge will be for both regulatory authorities and applicants, as well as for academia, to further elucidate the unresolved scientific questions.

Dose-response Studies on the Genotoxic Potential of Estragole and its Metabolite 1′-hydroxyestragole in Human Liver Cells

Due to its known genotoxicity and hepatotoxicity, estragole and structurally related phenylpropenes such as methyleugenol are of concern [96]. Recently, the HMPC has recommended a reduction in the contents of estragole in HMPs to below the guidance value of 0.05 mg per person and day [81]. Following its oral uptake, estragole is transported to the liver, where it undergoes extensive phase I metabolism to form different products. O-demethylation of estragole to 4-allylphenole (chavicol) is a detoxification step mediated by CYPs [97]. Its epoxidation to estragole-2′,3′-epoxide and subsequent hydrolysis to estragole-2′,3′-diole catalyzed by epoxide hydrolase are also considered as detoxification reactions [98]. Moreover, estragole can be bio-activated to 1′-hydroxyestragol, which is catalyzed by CYP1A2 and CYP2A6 [99]. The hydroxylated metabolite is then converted to 1′-sulphooxyestragole by sulfotransferase 1A1 (SULT1A1) and SULT1C2 [100], [101]. The formed intermediate spontaneously decomposes, giving rise to a reactive carbenium ion and subsequent DNA damage, with trans-E3′-N ²-1′-desoxyguanosine (E-3′-N2-dG) and E3′-N ⁶-2′-desoxyadenosine (E-3′-N6-dA) as the main DNA adducts [102], [103]. DNA adduct formation was recently shown to occur in a concentration- and time-dependent manner in primary rat hepatocytes, with maximum adduct levels after 6 h [104]. At later time points, the adduct levels moderately decreased, which may be attributable to DNA repair. In line with this notion, a recent study provided evidence for the involvement of DNA repair in the removal of E-3′-N2-dG adducts, albeit with limited efficiency [105]. Another explanation for the decreased adduct levels found at later time points could be cytotoxicity, which is indeed observed at high estragole concentrations [104]. As mentioned before, estragole is structurally closely related to methyleugenol, and both share a common bio-activation pathway. Interestingly, higher levels of methyleugenol-derived DNA adducts cause replication stress in liver cells, which triggers p53-dependent mitochondrial apoptosis [106]. The concentration-response studies with estragole in primary rat hepatocytes further indicated a point of departure (PoD) for DNA adduct formation [104]. Such nonlinear concentration-response curves can be caused by DNA repair pathways [107], as previously shown for DNA alkylation damage [108], [109], [110], [111]. With regard to human metabolically competent liver cell models, there are little quantitative genotoxicity data available so far for estragole. Therefore, current research investigates the concentration-dependent cytotoxicity and genotoxicity of estragole and its main phase I metabolite 1′-hydroxyestragole in human liver cell models, including genetically engineered HepG2 cells and primary human hepatocytes. The concentration-response data will be used for benchmark concentration modeling and hockey-stick modeling, in order to determine whether a ʼvirtually no effectʼ PoD for the genotoxic mode of action exists in human liver cells or not. These results are eagerly awaited and will hopefully contribute to refining the risk assessment for estragole in food and HMPs.

Summary and Conclusions

This chapter is aimed at summarizing the contributions and results from the workshop and presenting conclusions based on these results. A broader discussion, taking into account all relevant data and drawing overall conclusions from those, would have been beyond the scope of this report.

PAs are found in approximately 2 – 3% of all flowering plants worldwide. A major source of contamination of feed and food is the occurrence in weeds not sufficiently removed during the harvest of the cultured plants. Uptake from soil and subsequent distribution into various parts of the plant represents another pathway of contamination. The risk of adverse health effects in humans and livestock has triggered intense research on the occurrence, exposure, and risk assessment over the last decade. Studies on the transfer of PAs from farm animals to food products indicate, e.g., limited overall transfer of PAs into cowʼs milk or into eggs, although strong differences seem to prevail between congeners. In particular in the rumen, the reductive metabolism of PAs results in the conversion of PA N-oxides into their parent PAs. In insects raised for food or feed, e.g., in mealworm larvae, much higher transfer rates from the feed into the larvae were found for a number of congeners. Taken together, these data indicate that a transfer from farm animals into food occurs, usually to a low extent, depending, however, on the congener. For insects used as food or feed, substantially higher transfer rates of parent PAs are possible.

In order to assess the relative genotoxic potencies of PAs in vitro, the metabolically competent human HepaRG liver cell line using the γH2AX assay is a suitable tool. In combination with LC-MS/MS analysis, less-investigated congeners with a substantial genotoxicity can be identified. Furthermore, this methodology has been used successfully to address the issue of possible interactions between PAs and pesticides with respect to the genotoxic outcome. Recent data suggest that co-exposure in humans to PAs and DDT may enhance hepatotoxicity. In HepaRG cells, it was found that several pesticides can induce CYP3A4 with a subsequent increase in PA genotoxicity. This suggests that these pesticides (and other chemicals that induce CYP3A4) may potentiate the (geno)toxic effects of PAs upon co-exposure if sufficient internal concentrations for CYP induction are reached.

More research has been carried out with respect to the role of metabolism in PA toxicity. The three major pathways are hydrolysis of the ester group(s), N-oxidation to form PANOs, and hydroxylation of the necine bases followed by spontaneous dehydration to produce dehydro-PAs (DHPAs). The latter together with their hydrolytic product 1-hydroxymethyl-7-hydroxy-6,7-dihydropyrrolizine (DHP) are considered to be the reactive metabolites responsible for toxicity. In experimental animals, CYP3A and 2B subfamilies were the major CYPs catalyzing the metabolisms of several Pas, while in humans, CYP3A4 and 3A5 are critical in the activation of retronecine-type PAs, with the exception of monocrotaline being mainly bio-activated by CYP2A6. In lymphoblastoid TK6 cell lines expressing individual human CYPs, the dominant role of CYP3A4 in bio-activation was confirmed for the metabolism of prototype congeners from heliotridine-, retronecine-, and otonecine-type PAs. CYP3A5 and 3A7 were found to play a minor role. The results were confirmed in the micronucleus assay and in cell cycle analysis. A relative-potency analysis for genotoxicity revealed a broad spectrum of potencies with lasiocarpine being the most potent and lycopsamine the least potent in a row of 13 congeners. The relative potencies spanned several orders of magnitude and were in striking agreement with published results from other human and rat cell models. In particular, the low potencies of monoesters together with the cyclic diester monocrotaline were confirmed. The data indicate that relative potency estimates for a key endpoint in the risk assessment of PAs, i.e., for genotoxicity, are converging between in vitro studies making the basis for risk-assessment decisions more solid.

Likewise, in HepaRG cells, the genotoxic potencies of a broad spectrum of representative PAs differ significantly due to the different structures. Under standard in vitro conditions with ambient air (21%) oxygen levels, N-oxides are less potent inducers of DNA damage than their corresponding parent PA. With heliotrine, it was found that, under these conditions, the N-oxide was a less potent genotoxicant than under low oxygen (2%), whereas the opposite was found for the parent PA. Chemical analysis revealed that several-fold more potent PA is formed under low oxygen when compared to a standard ambient air oxygen concentration. These data indicate that the oxygen level has a significant impact on the genotoxic potency of heliotrine N-oxide, which should have a future impact on the design and interpretation of respective in vitro studies.

Since PANOs are often the predominant form of PAs in plants, their risk assessment is of special importance. Reduction into the parent PAs can occur via the intestinal microbiota but also via enzymes in the hostʼs tissues. Using PBK modeling based on in silico parameters and parameters derived from in vitro experiments, it is possible to predict the relative potency factors using toxicokinetic parameters for parent PAs vs. PANOs for both rats and humans. The predictions can then be compared to the amount of pyrrole–protein adducts as an effect marker. PBK-based predictions of toxicokinetics for riddelliine N-oxide, senecionine N-oxide, and their parent PAs in rats were close to in vivo-derived values. Furthermore, the models suggest that the relative potencies are dependent on dose, species, and endpoint. Modeling at low, relevant dose levels may allow the derivation of estimates relevant for risk assessment with an extrapolation to humans.

PBK models are usually based on in vitro data on metabolism, DNA adduct formation, etc., together with measures of oral absorption and metabolic bio-activation relevant to the gastrointestinal tract. Similar efforts are warranted considering the dermal route of exposure and the extent to which PAs could be absorbed and/or metabolized in the skin. The ratio of DHP-DNA adducts/in vitro AUC (determined in rat hepatocyte cultures) was found to be correlated to the intrinsic toxic potency of congeners. Furthermore, both predicted and experimental lipophilicity parameters were correlated with DNA adduct/AUC values. On this basis, it can be shown that LogD values are useful to predict DNA adduct/AUC values and relative potency estimates for PAs with insufficient data. Even under the assumption of complete dermal bioavailability, these considerations result in margin-of-exposure estimates for topically applied HMCs well above 10 000, suggesting a low concern for human health. Furthermore, actual measurements with in vitro models for dermal absorption revealed low dermal penetration values of < 10% of the applied dose for lycopsamine, echimidine, or retrorsine. Incubations with human skin S9 preparations did not show any metabolic losses with lycopsamine, intermedine, 7-acetylintermedine, 7-acetyllycopsamine, or echimidine, suggesting marginal or negligible metabolic activation in the skin.

Since PAs may be found in plant-derived herbal medicinal products (HMPs), there is a requirement for adequate action by pharmaceutical companies and regulatory authorities. These requirements have been addressed, e.g., by European authorities, which published mitigation plans and public statements including recommendations for restrictions and requirements. The latter have been adapted to the evolving scientific knowledge. Examples are requirements for specification limits, analytical methods, and their validation. A current challenge is the introduction of adjusted specification limits for PAs in HMPs for children. A further improvement in the control of PA contaminations in HMPs can be expected from future guidance documents and the progress made by agricultural producers, pharmaceutical companies, and test laboratories in the implementation of relevant measures.

The fact that pyrrolic PA metabolites can bind to proteins, thus forming detectable adducts, offers a possibility to analyze these as a biomarker not only of exposure but also of effect. For example, this method allowed the study of the role of the gut–liver interplay in PA toxicity. Upon long-term treatment of mice with retrorsine, the intestinal mucosa can be damaged with a subsequent disruption of the intestinal barrier. When intestinal CYP activity is suppressed, the formation of pyrrolic metabolites was not affected substantially, however, indicating that hepatic metabolism and biliary excretion of metabolites but not intestinal metabolic activation are mainly responsible for the observed enterotoxicity. When chronic colitis was induced with DSS, the hepatotoxicity of retrorsine was enhanced. Although CYP activity in the liver was unaltered under these conditions, depletion of hepatic glutathione was observed. Furthermore, damage of the intestinal barrier upon colitis may enhance endotoxin transport to the liver and may also contribute to the enhanced liver damage observed. These findings illustrate the role of the intact intestine in PA toxicity in the liver and suggest a possibly higher susceptibility to PAs in patients with intestinal affections.

An increasing number of publications provide convincing evidence that the toxicity of PAs differs considerably between congeners. Establishing relative potency (REP) factors would be an important tool for the refinement of the risk assessment of complex PA mixtures. Such mixtures are usually present in plants and, consequently, in food, feed, and herbal preparations. Their current assessment is based on the assumption that all 1,2-unsaturated PAs are as toxic as the most toxic congeners(s), i.e., lasiocarpine or riddelliine, and thus may lead to an inadequate overestimation of the risk.

Recent in vitro metabolism studies in human and rat liver microsomes, in rat hepatocytes, and in CYP3A4-expressing human HepG2 cells using a broad spectrum of PA congeners revealed that their relative toxic potency is related to the rate and extent of the metabolism of the parent PA. Highly potent congeners were extensively metabolized, whereas several monoesters with low toxic potency were not. A detailed analysis of the GSH adducts in human liver microsomes revealed that those extensively metabolized, highly toxic PAs also form detectable amounts of one or more GSH adducts, while the monoesters did not. Thus, extensive metabolism and formation of detectable amounts of GSH conjugates are obviously markers for high toxic potency. Since relative toxic potencies in intact cells for a few congeners were in accordance with the measured GSH conjugate formation in a subcellular metabolically active fraction (liver microsomes), differences in uptake and distribution seem to play a limited role for the relative toxic potency in these cases. Further studies on this issue are warranted.

The concept of iREPs was supported by a number of recent publications on in vitro models of both human and rodent origin. The structure–potency relationship was also in very good agreement with the published concept with the exception of monocrotaline. In vitro human metabolism data confirm, however, that monocrotaline is a weak substrate for microsomal metabolism and does not form detectable amounts of GSH conjugates. This exception is likely due to the peculiar ring structure of monocrotaline.

Alkenylbenzenes are secondary plant metabolites that occur in various herbs and spices, like basil, fennel, and parsley. Exposure to alkenylbenzenes is a result of a direct consumption of these herbs and spices, as well as of the use of essential oils as flavoring ingredients or the use of herbal supplements. In addition, alkenylbenzene-containing essential oils are used more and more as feed additives. Given the potential genotoxic and carcinogenic effect of various alkenylbenzenes, the presence of these substances in food and feed poses a health concern. The fact that only a few rodent studies at relatively high dose levels are available makes the risk assessment of these compounds difficult. Furthermore, no data are available on their possible carcinogenicity in humans. The toxicological data focus on estragole, safrole, or methyleugenol, while much less information is available for other alkenylbenzenes.

Upon integration of toxicokinetic (transport and metabolism) data in a PBK model, one can simulate the bio-activation and detoxification in a species of interest. This approach, for example, allowed the determination that dose-dependent effects and species differences in bio-kinetics cannot be used as arguments to reduce default uncertainty factors in the risk evaluation of alkenylbenzenes.

Also, differences in the extent of bio-activation between different alkenylbenzenes could be established. Furthermore, the flavonoid nevadensin was identified to inhibit sulfotransferase (SULT)-mediated bioactivation and DNA adduct formation. However, these matrix-derived interactions appear to be dependent and predicted by the PBK models to be absent at realistic low dietary human intake. Overall, in vitro kinetic experiments in combination with PBK modeling provide a relevant framework to explore dose-dependent effects, species (human vs. animal) differences, human variation, and matrix modulation of alkenylbenzenes.

Alkenylbenzenes occur in a large number of plant families, e.g., Apiaceae, Asteraceae, Rosaceae, Solanaceae, Piperaceae, Rutaceae, or Poaceae. Because of their volatility, they are mainly found in the respective essential oils. However, alkenylbenzenes may also occur in the form of nonvolatile glycosides. Because of their favorable sensory properties, pure alkenylbenzenes or essential oils containing them have been used in industrial food production for decades. The increasing knowledge of the genotoxic and carcinogenic effects prompted regulators in the USA to prohibit the use of safrole, beta-asarone, and methyleugenol. In the EU, the use of pure safrole, asarone, methyleugenol, and estragole in food has been prohibited since 2011. Permissible concentrations in certain food categories are defined for these alkenylbenzenes naturally present in flavorings and food ingredients with flavoring properties and in certain compound foods, as consumed, to which flavorings and/or food ingredients with flavoring properties have been added. The HMPC adopted additional statements on the use of methyleugenol and estragole in HMP in 2005. At that time, with reference to interspecies differences of metabolic activation steps, the probable nonlinear dose-response curve at low doses, and the low doses and short treatment duration typically found in the medicinal use of relevant herbal products, no dose limits were imposed, but the committee stipulated the need for further data, including data on background exposure levels from food intake. On 1 March 2022, the HMPC published a revised public statement on the use of HMPs containing estragole. This document sets a “guidance value” of 1 µg/kg bw as a maximum daily dose for children and pregnant and breastfeeding women and demands reducing the estragole content in other HMPs as much as practicable. The necessity, with reference to the ICH M7 guideline, to consider the duration of use and background exposure from food is stipulated.

In 2022, the HMPC drafted revisions of the monographs on bitter fennel fruit and sweet fennel fruit, as well as of bitter fennel oil. The HMPC no longer endorses the use of bitter fennel oil in herbal medicinal products due to the high dosage specified in the hitherto existing HMPC monograph and the resulting high estragole intake. It will be on the part of marketing authorization/registration holders to elaborate on whether other measures are reasonably practicable to further reduce the content of estragole in their products. Although these documents continue to principally allow the use of estragole-containing HMPs and fennel fruit in children, it will be a challenge for many existing products to meet the guidance value since robust quantitative data on background exposure to estragole via food are not available today.

Upon oral uptake, estragole is transported to the liver, where it undergoes extensive phase I metabolism. This includes detoxification steps such as O-demethylation or epoxidation to estragole-2′,3′-epoxide and subsequent hydrolysis to estragole-2′,3′-diole. Estragole can also be bioactivated to 1′-hydroxyestragol and catalyzed by CYP1A2 and CYP2A6. The hydroxylated metabolite is then converted to 1′-sulphooxyestragole by sulfotransferases and may spontaneously decompose, giving rise to a reactive carbenium ion. The latter can form various DNA adducts in a concentration- and time-dependent manner. Higher levels of methyleugenol-derived DNA adducts cause replication stress in liver cells, which triggers p53-dependent mitochondrial apoptosis. Concentration-response studies on DNA adduct levels with estragole in primary rat hepatocytes further indicated a point of departure (PoD) for DNA adduct formation. Such nonlinear concentration-response curves can be caused, e.g., by detoxification mechanisms and DNA repair pathways, as previously shown for DNA alkylation damage. Concentration-response modeling is warranted in order to determine whether a ʼvirtually no effectʼ PoD for the genotoxic mode of action exists in human liver cells or not.

Disclaimer

This article reflects the views of the authors and should not be construed to represent the views or policies of the U. S. Food and Drug Administration (FDA). Any mention of commercial products, their sources, or their use in connection with the material reported herein is for clarification only and is not to be construed as either an actual or implied endorsement of such products.

Contributorsʼ Statement

All authors:

• supervision of generation of data and/or concepts presented/discussed in the individual chapters

• design of the concept of the paper

• drafting of chapters

• drawing of figures (if appropriate)

• designing of tables (if appropriate)

• response to reviewers

• final reading

Dieter Schrenk:

• all of the above-mentioned plus coordination of the drafting process

Conflict of Interest

The authors declare that they have no conflict of interest.

Acknowledgements

The authors want to acknowledge the work of their colleagues and collaborators who have contributed substantially to the original work (which has been published or is in the publication process) discussed in this review: Gabriel Ackermann, Yasser Alhejji, Marc Audebert, Greg Dameron, Mona Delagrange, Andrea Edginton, Peter Fu, Lan Gao, Ina Geburek, Arjen Gerssen, Manuel Haas, Yesheng He, Wendy Jansen Holleboom, Mike Karb, James Keating, Jan-Heiner Küpper, Cathy Lester, Jochem Louisse, Jiang Ma, Nan Mai, Karl-Heinz Merz, Cindy Obringer, Ivonne Rietjens, Deborah Rijkers, Lukas Rutz, Milou van de Schans, Bernd Schäfer, Geert Stoopen, Anja These, Sebastiaan Wesseling, Jamyang Yangchen, Liang Zheng, and Rubing Zou. The help of Arie van der Bent and Deborah Rijkers with Figures 3 and 4 is gratefully acknowledged.

• References

• 1 EFSA (European Food Safety Authority). Scientific Panel on Contaminants in the Food Chain (CONTAM). Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions and food supplements. EFSA J 2017; 15: 4908
• 2 Bodi D, Ronczka S, Gottschalk C, Behr N, Skibba A, Wagner M, Lahrssen-Wiederholt M, Preiss-Weigert A, These A. Determination of pyrrolizidine alkaloids in tea, herbal drugs and honey. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2014; 31: 1886-1895
• 3 Schrenk D, Fahrer J, Allemang A, Fu P, Lin G, Mahony C, Mulder PPJ, Peijnenburg A, Pfuhler S, Rietjens IMCM, Sachse B, Steinhoff B, These A, Troutman J, Wiesner J. Novel insights into pyrrolizidine alkaloid toxicity and implications for risk assessment: occurrence, genotoxicity, toxicokinetics, risk assessment-a workshop report. Planta Med 2022; 88: 98-117
• 4 Clayton MJ, Davis TZ, Knoppel EL, Stegelmeier BL. Hepatotoxic plants that poison livestock. Vet Clin North Am Food Anim Pract 2020; 36: 715-723
• 5 Jiao W, Wang L, Zhu L, Shen T, Shi T, Zhang P, Wang C, Chen H, Wu X, Yang T, Li Q, Hua R. Pyrrolizidine-producing weeds in tea gardens as an indicator of alkaloids in tea. Food Addit Contam Part B Surveill 2023; 16: 50-57
• 6 Fu PP. Pyrrolizidine alkaloids: Metabolic activation pathways leading to liver tumor initiation. Chem Res Toxicol 2017; 30: 81-93
• 7 Neuman MG, Cohen L, Opris M, Nanau RM, Hyunjin J. Hepatotoxicity of pyrrolizidine alkaloids. J Pharm Pharm Sci 2015; 18: 825-843
• 8 Datta DV, Khuroo MS, Mattocks AR, Aikat BK, Chhuttani PN. Veno-occlusive disease of liver due to heliotropium plant, used as medicinal herb (report of 6 cases with review of literature). J Assoc Physicians India 1978; 26: 383-393
• 9 Yang M, Ma J, Ruan J, Ye Y, Fu PPC, Lin G. Intestinal and hepatic biotransformation of pyrrolizidine alkaloid N-oxides to toxic pyrrolizidine alkaloids. Arch Toxicol 2019; 93: 2197-2209
• 10 Chan P. NTP technical report on the toxicity studies of riddelliine (CAS No. 23246–96-0) administered by gavage to F344 rats and B6C3F1 mice. Toxic Rep Ser 1993; 27: 1-D9
• 11 Merz KH, Schrenk D. Interim relative potency factors for the toxicological risk assessment of pyrrolizidine alkaloids in food and herbal medicines. Toxicol Lett 2016; 263: 44-57
• 12 EFSA Panel on Contaminants in the Food Chain (CONTAM). Scientific opinion on pyrrolizidine alkaloids in food and feed. EFSA J 2011; 9: 2406
• 13 Mulder PPJ, López P, Castellari M, Bodi D, Ronczka S, Preiss-Weigert A, These A. Occurrence of pyrrolizidine alkaloids in animal- and plant-derived food: Results of a survey across Europe. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2018; 35: 118-133
• 14 Hoogenboom RLAP, Mulder PPJ, Zeilmaker MJ, van den Top HJ, Remmelink GJ, Brandon EFA, Klijnstra M, Meijer GAL, Schothorst R, Van Egmond HP. Carry-over of pyrrolizidine alkaloids from feed to milk in dairy cows. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2011; 28: 359-372
• 15 Mulder PPJ, Klijnstra MD, Goselink RMA, van Vuuren AM, Cone JW, Stoopen G, Hoogenboom RLAP. Transfer of pyrrolizidine alkaloids from ragwort, common groundsel and viperʼs bugloss to milk from dairy cows. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2020; 37: 1906-1921
• 16 Lamp J, Knappstein K, Walte HG, Krause T, Steinberg P, Schwake-Anduschus C. Transfer of tropane alkaloids (atropine and scopolamine) into the milk of subclinically exposed dairy cows. Food Control 2021; 126: 108056
• 17 Engel AM, Klevenhusen F, Moenning JL, Numata J, Fischer-Tenhagen C, Sachse B, Schäfer B, Fry H, Kappenstein O, Pieper R. Investigations on the transfer of quinolizidine alkaloids from Lupinus angustifolius into the milk of dairy cows. J Agric Food Chem 2022; 70: 11749-11758
• 18 Mulder PPJ, de Witte SL, Stoopen GM, van der Meulen J, van Wikselaar PG, Gruys E, Groot MJ, Hoogenboom RLAP. Transfer of pyrrolizidine alkaloids from various herbs to eggs and meat in laying hens. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2016; 33: 1826-1839
• 19 Mulder PPJ, Mueller-Maatsch JTL, Meijer N, Bosch M, Zoet L, van der Fels-Klerx HJ. Effects of dietary exposure to plant toxins on bioaccumulation, survival, and growth of black soldier fly (Hermetia illucens) larvae and lesser mealworm (Alphitobius diaperinus). Heliyon 2023; DOI: 10.2139/ssrn.4498727. Jul 19 [Epub ahead of Print]
• 20 Kaltner F, Rychlik M, Gareis M, Gottschalk C. Influence of storage on the stability of toxic pyrrolizidine alkaloids and their N-oxides in peppermint tea, hay, and honey. J Agric Food Chem 2018; 66: 5221-5228
• 21 de Nijs M, Mulder PPJ, Klijnstra MD, Driehuis F, Hoogenboom RLAP. Fate of pyrrolizidine alkaloids during processing of milk of cows treated with ragwort. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2017; 34: 2212-2219
• 22 Fu PP, Xia Q, Lin G, Chou MW. Pyrrolizidine alkaloids–genotoxicity, metabolism enzymes, metabolic activation, and mechanisms. Drug Metab Rev 2004; 36: 1-55
• 23 Ruan J, Yang M, Fu P, Ye Y, Lin G. Metabolic activation of pyrrolizidine alkaloids: Insights into the structural and enzymatic basis. Chem Res Toxicol 2014; 27: 1030-1039
• 24 Ebmeyer J, Braeuning A, Glatt H, These A, Hessel-Pras S, Lampen A. Human CYP3A4-mediated toxification of the pyrrolizidine alkaloid lasiocarpine. Food Chem Toxicol 2019; 130: 79-88
• 25 Rutz L, Gao L, Küpper JH, Schrenk D. Structure-dependent genotoxic potencies of selected pyrrolizidine alkaloids in metabolically competent HepG2 cells. Arch Toxicol 2020; 94: 4159-4172
• 26 Li X, He X, Chen S, Guo X, Bryant MS, Guo L, Manjanatha MG, Zhou T, Witt KL, Mei N. Evaluation of pyrrolizidine alkaloid-induced genotoxicity using metabolically competent TK6 cell lines. Food Chem Toxicol 2020; 145: 111662
• 27 Li X, Chen S, Guo X, Wu Q, Seo JE, Guo L, Manjanatha MG, Zhou T, Witt KL, Mei N. Development and application of TK6-derived cells expressing human cytochrome P450 s for genotoxicity testing. Toxicol Sci 2020; 175: 251-265
• 28 Allemang A, Mahony C, Lester C, Pfuhler S. Relative potency of fifteen pyrrolizidine alkaloids to induce DNA damage as measured by micronucleus induction in HepaRG human liver cells. Food Chem Toxicol 2018; 121: 72-81
• 29 Louisse J, Rijkers D, Stoopen G, Holleboom WJ, Delagrange M, Molthof E, Mulder PPJ, Hoogenboom RLAP, Audebert M, Peijnenburg AACM. Determination of genotoxic potencies of pyrrolizidine alkaloids in HepaRG cells using the γH2AX assay. Food Chem Toxicol 2019; 131: 110532
• 30 Louisse J, Mulder PPJ, Gerssen A, Stoopen G, Rijkers D, van de Schans MGM, Peijnenburg AACM. Bioassay-directed analysis-based identification of relevant pyrrolizidine alkaloids. Arch Toxicol 2022; 96: 2299-2317
• 31 Robinson O, Want E, Coen M, Kennedy R, van den Bosch C, Gebrehawaria Y, Kudo H, Sadiq F, Goldin RD, Hauser ML, Fenwick A, Toledano MB, Thursz MR. Hirmi Valley liver disease: A disease associated with exposure to pyrrolizidine alkaloids and DDT. J Hepatol 2014; 60: 96-102
• 32 Yang M, Ruan J, Gao H, Li N, Ma J, Xue J, Ye Y, Fu PP, Wang J, Lin G. First evidence of pyrrolizidine alkaloid N-oxide-induced hepatic sinusoidal obstruction syndrome in humans. Arch Toxicol 2017; 91: 3913-3925
• 33 He X, Xia Q, Woodling K, Lin G, Fu PP. Pyrrolizidine alkaloid-derived DNA adducts are common toxicological biomarkers of pyrrolizidine alkaloid N-oxides. J Food Drug Anal 2017; 25: 984-991
• 34 Yang M, Ma J, Ruan J, Zhang C, Ye Y, Fu PPC, Lin G. Absorption difference between hepatotoxic pyrrolizidine alkaloids and their N-oxides–Mechanism and its potential toxic impact. J Ethnopharmacol 2020; 249: 112421
• 35 Widjaja F, Alhejji Y, Rietjens IM. The role of kinetics as key determinant in toxicity of pyrrolizidine alkaloids and their N-oxides. Planta Med 2022; 88: 130-143
• 36 Widjaja F, Wesseling S, Rietjens IMCM. Physiologically based kinetic modelling predicts the in vivo relative potency of riddelliine N-oxide compared to riddelliine in rat to be dose dependent. Arch Toxicol 2022; 96: 135-151
• 37 Widjaja F, Alhejji Y, Yangchen J, Wesseling S, Rietjens IMCM. Physiologically-based kinetic modeling predicts similar in vivo relative potency of senecionine N-oxide for rat and human at realistic low exposure levels. Mol Nutr Food Res 2023; 67: e2200293
• 38 Widjaja F, Zheng L, Wesseling S, Rietjens IMCM. Physiologically based kinetic modeling of senecionine N-oxide in rats as a new approach methodology to define the effects of dose and endpoint used on relative potency values of pyrrolizidine alkaloid N-oxides. Front Pharmacol 2023; 14: 1125146
• 39 Alhejji Y, Widjaja F, Tian S, Hoekstra T, Wesseling S, Rietjens IM. The influence of dose, fraction bioactivated and endpoint used on the relative potency value of selected PA N-oxides compared to the parent PAs. Curr Res Toxic (submitted).
• 40 Williams L, Chou MW, Yan J, Young JF, Chan PC, Doerge DR. Toxicokinetics of riddelliine, a carcinogenic pyrrolizidine alkaloid, and metabolites in rats and mice. Toxicol Appl Pharmacol 2002; 182: 98-104
• 41 Xia Q, Zhao Y, Von Tungeln LS, Doerge DR, Lin G, Cai L, Fu PP. Pyrrolizidine alkaloid-derived DNA adducts as a common biological biomarker of pyrrolizidine alkaloid-induced tumorigenicity. Chem Res Toxicol 2013; 26: 1384-1396
• 42 Xia Q, Zhao Y, Lin G, Beland FA, Cai L, Fu PP. Pyrrolizidine alkaloid-protein adducts: Potential non-invasive biomarkers of pyrrolizidine alkaloid-induced liver toxicity and exposure. Chem Res Toxicol 2016; 29: 1282-1292
• 43 Lester C, Troutman J, Obringer C, Wehmeyer K, Stoffolano P, Karb M, Xu Y, Roe A, Carr G, Blackburn K, Mahony C. Intrinsic relative potency of a series of pyrrolizidine alkaloids characterized by rate and extent of metabolism. Food Chem Toxicol 2019; 131: 110523
• 44 Martinez MN, Amidon GL. A mechanistic approach to understanding the factors affecting drug absorption: A review of fundamentals. J Clin Pharmacol 2002; 42: 620-643
• 45 Plaza A, Toner F, Harris J, Ottersbach P, Roper C, Mahony C. Support for regulatory assessment of percutaneous absorption of retronecine-type pyrrolizidine alkaloids through human skin. Planta Med 2022; 88: 144-151
• 46 EFSA (European Food Safety Authority). Guidance on dermal absorption. EFSA J 2017; 15: 4873
• 47 OECD (Organisation for Economic Co-operation and Development). Test No. 428: Skin Absorption: In Vitro Method. OECD Guidelines for the Testing of Chemicals, Section 4. Paris: OECD Publishing; 2004
• 48 OECD (Organisation for Economic Co-operation and Development). Guidance Document for the Conduct of Skin Absorption Studies, OECD Series on Testing and Assessment, No. 28. Paris: OECD Publishing,; 2004
• 49 SCCS (Scientific Committee on Consumer Safety). Basic criteria for the assessment of dermal absorption of cosmetic ingredients, 22 June 2010. Accessed May 13, 2023 at: https://ec.europa.eu/health/scientific_committees/consumer_safety/docs/sccs_s_002.pdf
• 50 HMPC (Committee on Herbal Medicinal Products). Public statement on the use of herbal medicinal products containing toxic, unsaturated pyrrolizidine alkaloids (PAs), EMA/HMPC/893108/2011 Rev. 1, 07.07.2021. Accessed May 04, 2023 at: https://www.ema.europa.eu/en/documents/public-statement/public-statement-use-herbal-medicinal-products-containing-toxic-unsaturated-pyrrolizidine-alkaloids-pas-including-recommendations-regarding-contamination-herbal-medicinal-products-pyrrolizidine_en.pdf
• 51 BfArM (Bundesinstitut für Arzneimittel und Medizinprodukte). Bekanntmachung zur Prüfung des Gehalts an Pyrrolizidinalkaloiden zur Sicherstellung der Qualität und Unbedenklichkeit von Arzneimitteln, die pflanzliche Stoffe bzw. pflanzliche Zubereitungen oder homöopathische Zubereitungen aus pflanzlichen Ausgangsstoffen als Wirkstoffe enthalten. BfArM Bekanntmachung 01.03.2016 (in German). Accessed April 26, 2023 at: https://www.bfarm.de/SharedDocs/Bekanntmachungen/DE/Arzneimittel/besTherap/bm-besTherap-20160301-pa-pdf.html
• 52 BfR (Bundesinstitut für Risikobewertung). Aktualisierte Risikobewertung zu Gehalten an 1, 2-ungesättigten Pyrrolizidinalkaloiden (PA) in Lebensmitteln Juni 2018 (Stellungnahme Nr. 020/2018) 06.2018 (in German). Accessed May 13, 2023 at: https://www.bfr.bund.de/cm/343/aktualisierte-risikobewertung-zu-gehalten-an-1-2-ungesaettigten-pyrrolizidinalkaloiden-pa-in-lebensmitteln.pdf
• 53 BfR (Bundesinstitut für Risikobewertung). Bestimmung von Pyrrolizidinalkaloiden (PA) in Pflanzenmaterial mittels SPE-LC-MS/MS (BfR-PA-Tee-2.0/2014) 10.2014 (in German). Accessed May 27, 2023 at: https://www.bfr.bund.de/cm/343/bestimmung-von-pyrrolizidinalkaloiden.pdf
• 54 European Pharmacopoeia. Vol. 10.6. Contaminant pyrrolizidine alkaloids, Council of Europe 01/2022: Ph. Eur. 2.8.26. Accessed December 07, 2023 at: https://pheur.edqm.eu/home
• 55 Lin G, Wang JY, Li N, Li M, Gao H, Ji Y, Zhang F, Wang H, Zhou Y, Ye Y, Hong XX, Zheng J. Hepatic sinusoidal obstruction syndrome associated with consumption of Gynura segetum. J Hepatol 2011; 54: 666-673
• 56 Mohabbat O, Younos MS, Merzad AA, Srivastava RN, Sediq GG, Aram GN. An outbreak of hepatic veno-occlusive disease in north-western Afghanistan. The Lancet 1976; 308: 269-271
• 57 He Y, Long Y, Zhang C, Ma J, Ke C, Tang C, Ye Y, Lin G. Dietary alcohol exacerbates the hepatotoxicity induced by pyrrolizidine alkaloids: Hazard from food contamination. J Hazard Mater 2022; 424: 127706
• 58 He Y, Zhu L, Ma J, Lin G. Metabolism-mediated cytotoxicity and genotoxicity of pyrrolizidine alkaloids. Arch Toxicol 2021; 95: 1917-1942
• 59 He Y, Lian W, Ding L, Fan X, Ma J, Zhang QY, Ding X, Lin G. Lung injury induced by pyrrolizidine alkaloids depends on metabolism by hepatic cytochrome P450 s and blood transport of reactive metabolites. Arch Toxicol 2021; 95: 103-116
• 60 He Y, Ma J, Fan X, Ding L, Ding X, Zhang QY, Lin G. The key role of gut–liver axis in pyrrolizidine alkaloid-induced hepatotoxicity and enterotoxicity. Acta Pharm Sin B 2021; 11: 3820-3835
• 61 Chen L, Mulder PPJ, Louisse J, Peijnenburg A, Wesseling S, Rietjens I. Risk assessment for pyrrolizidine alkaloids detected in (herbal) teas and plant food supplements. Regul Toxicol Pharmacol 2017; 86: 292-302
• 62 Schrenk D, Gao L, Lin G, Mahony C, Mulder PPJ, Peijnenburg A, Pfuhler S, Rietjens IMCM, Rutz L, Steinhoff B, These A. Pyrrolizidine alkaloids in food and phytomedicine: Occurrence, exposure, toxicity, mechanisms, and risk assessment – A review. Food Chem Toxicol 2020; 136: 111107
• 63 Herzog N, Katzenberger N, Martin F, Schmidtke KU, Küpper JH. Generation of cytochrome P450 3A4-overexpressing HepG2 cell clones for standardization of hepatocellular testosterone 6β-hydroxylation activity. J Cell Biotechnol 2015; 1: 15-26
• 64 Haas M, Wirachowski K, Thibol L, Kupper JH, Schrenk D, Fahrer J. Potency ranking of pyrrolizidine alkaloids in metabolically competent human liver cancer cells and primary human hepatocytes using a genotoxicity test battery. Arch Toxicol 2023; 97: 1413-1428
• 65 Geburek I, Schrenk D, These A. In vitro biotransformation of pyrrolizidine alkaloids in different species: Part II-identification and quantitative assessment of the metabolite profile of six structurally different pyrrolizidine alkaloids. Arch Toxicol 2020; 94: 3759-3774
• 66 Geburek I, Preiss-Weigert A, Lahrssen-Wiederholt M, Schrenk D, These A. In vitro metabolism of pyrrolizidine alkaloids – Metabolic degradation and GSH conjugate formation of different structure types. Food Chem Toxicol 2020; 135: 110868
• 67 Geburek I, Rutz L, Gao L, Küpper JH, These A, Schrenk D. Metabolic pattern of hepatotoxic pyrrolizidine alkaloids in liver cells. Chem Res Toxicol 2021; 34: 1101-1113
• 68 Shimshoni JA, Barel S, Mulder PPJ. Comparative risk assessment of three native Heliotropium species in Israel. Molecules 2021; 26: 689
• 69 Chen L, Zhang Q, Yi Z, Chen Y, Xiao W, Su D, Shi W. Risk assessment of (herbal) teas containing pyrrolizidine alkaloids (PAs) based on margin of exposure approach and Relative Potency (REP) Factors. Foods 2022; 11: 2946
• 70 Al-Subeihi AAA, Spenkelink B, Punt A, Boersma MG, van Bladeren PJ, Rietjens IMCM. Physiologically based kinetic modeling of bioactivation and detoxification of the alkenylbenzene methyleugenol in human as compared with rat. Toxicol Appl Pharmacol 2012; 260: 271-284
• 71 Martati E, Boersma MG, Spenkelink A, Khadka DB, van Bladeren PJ, Rietjens IMCM, Punt A. Physiologically based biokinetic (PBBK) modeling of safrole bioactivation and detoxification in humans as compared with rats. Toxicol Sci 2012; 128: 301-316
• 72 Punt A, Paini A, Boersma MG, Freidig AP, Delatour T, Scholz G, Schilter B, van Bladeren P, Rietjens IMCM. Use of physiologically based biokinetic (PBBK) modeling to study estragole bioactivation and detoxification in humans as compared with male rats. Toxicol Sci 2009; 110: 255-269
• 73 van den Berg SJPL, Punt A, Soffers AEMF, Vervoort J, Ngeleja S, Spenkelink B, Rietjens IMCM. Physiologically based kinetic models for the alkenylbenzene elemicin in rat and human and possible implications for risk assessment. Chem Res Toxicol 2012; 25: 2352-2367
• 74 Alhusainy W, van den Berg SJPL, Paini A, Campana A, Asselman M, Spenkelink A, Punt A, Scholz G, Schilter B, Adams TB, van Bladeren PJ, Rietjens IMCM. Matrix modulation of the bioactivation of estragole by constituents of different alkenylbenzene-containing herbs and spices and physiologically based biokinetic modeling of possible in vivo effects. Toxicol Sci 2012; 129: 174-187
• 75 Alhusainy W, Paini A, van den Berg JHJ, Punt A, Scholz G, Schilter B. In vivo validation and physiologically based biokinetic modeling of the inhibition of SULT-mediated estragole DNA adduct formation in the liver of male Sprague-Dawley rats by the basil flavonoid nevadensin. Mol Nutr Food Res 2013; 57: 1969-1978
• 76 Rietjens IM, Cohen SM, Fukushima S, Gooderham NJ, Hecht S, Marnett LJ, Smith RL, Adams TB, Bastaki M, Harman CG, Taylor SV. Impact of structural and metabolic variations on the toxicity and carcinogenicity of hydroxy- and alkoxy-substituted allyl and propenylbenzenes. Chem Res Toxicol 2014; 27: 1092-1103
• 77 Eisenreich A, Götz ME, Sachse B, Monien BH, Herrmann K, Schäfer B. Alkenylbenzenes in foods: Aspects impeding the evaluation of adverse health effects. Foods 2021; 10: 2139
• 78 Homburger F, Kelley jr. T, Friedler G, Russfield AB. Toxic and possible carcinogenic effects of 4-allyl-1, 2- methylenedioxybenzene (safrole) in rats on deficient diets. Med Exp Int J Exp Med 1961; 4: 1-11
• 79 EC (European Commission). Regulation No 1334/2008 of the European Parliament and of the Council of 16 December 2008 on flavourings and certain food ingredients with flavouring properties for use in and on foods and amending Council Regulation (EEC) No 1601/91, Regulations (EC) No 2232/96 and (EC) No 110/2008 and Directive 2000/13/EC. Off J Europ Union 31.12.2008. Accessed May 12, 2023 at: https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:354:0034:0050:en:PDF
• 80 SCF (Scientific Committee on Food). Opinion of the Scientific Committee on Food on estragole (1-allyl-4-methoxybenzene). SCF/CS/FLAV/FLAVOUR/6 ADD 2 Final 26 September 2001. Accessed April 04, 2023 at: https://ec.europa.eu/food/system/files/2016-10/fs_food-improvement-agents_flavourings-out104.pdf
• 81 EMA (European Medicines Agency). Committee on Herbal Medicinal Products (HMPC). Public statement on the use of herbal medicinal products containing estragole. EMA/HMPC/137212/2005 Rev 1, 22 September 2021. (Publication March 1st, 2022). Accessed December 07, 2023 at: https://www.ema.europa.eu/en/documents/herbal-comments/overview-comments-received-draft-revised-public-statement-use-herbal-medicinal-products-containing-estragole-emahmpc1372122005-rev-1-2nd-draft_en.pdf
• 82 EMEA (European Medicines Agency). Herbal Medicinal Products Working Party (HMPWP): Position paper on the use of herbal medicinal products containing asarone, 2003. EMEA/HMPWP/41/03 Final. Accessed March 16, 2023 at: https://www.ema.europa.eu/en/documents/press-release/meeting-emea-working-party-herbal-medicinal-products-30-june-and-1-july-2003_en.pdf
• 83 EMEA (European Medicines Agency). Herbal Medicinal Products Committee (HMPC). Public statement on the use of herbal medicinal products containing asarone, 2005. EMEA/HMPC/139215/2005. Accessed May 29, 2023 at: https://www.ema.europa.eu/en/documents/scientific-guideline/public-statement-use-herbal-medicinal-products-containing-methyleugenol_en.pdf
• 84 EMEA (European Medicines Agency). Herbal Medicinal Products Committee (HMPC). Public Statement on the use of herbal medicinal products containing methyleugenol, 2005. EMEA/HMPC/138363/2005. Accessed May 12, 2023 at: https://www.ema.europa.eu/en/documents/scientific-guideline/public-statement-use-herbal-medicinal-products-containing-methyleugenol_en.pdf
• 85 EMEA (European Medicines Agency). Herbal Medicinal Products Committee (HMPC). Public Statement on the use of herbal medicinal products containing estragole, 2005. EMEA/HMPC/137212/2005. Accessed April 25, 2023 at: https://www.ema.europa.eu/en/documents/scientific-guideline/public-statement-use-herbal-medicinal-products-containing-estragole_en.pdf
• 86 Bristol DW. NTP 3-month toxicity studies of estragole (CAS No. 140–67-0) administered by gavage to F344/N rats and B6C3F1 mice. Toxic Rep Ser 2011; 82: 1-111
• 87 EFSA Scientific Cooperation (ESCO) Report. Advice on the guidance document for the safety assessment of botanicals and botanical preparations intended for use as food supplements, based on real case studies. EFSA J 2009; 7: 280
• 88 EMA (European Medicines Agency). Committee on Herbal Medicinal Products (HMPC) Public statement on the use of herbal medicinal products containing estragole. EMA/HMPC/137212/2005 Rev 1 (draft published 2014). Accessed December 07, 2023 at: https://www.ema.europa.eu/en/documents/public-statement/public-statement-use-herbal-medicinal-products-containing-estragole_en.pdf
• 89 EMA (European Medicines Agency). Committee for Human Medicinal Products. ICH guideline M7(R1) on assessment and control of DNA reactive (mutagenic) impurities in pharmaceuticals to limit potential carcinogenic risk – Step 5 (EMA/CHMP/ICH/83812/2013), 25 August 2015. Accessed December 07, 2023 at: https://www.ema.europa.eu/en/documents/scientific-guideline/ich-guideline-m7r1-assessment-and-control-dna-reactive-mutagenic-impurities-pharmaceuticals-limit-potential-carcinogenic-risk-step-5_en.pdf
• 90 EMA (European Medicines Agency). Co-ordination Group for Mutual Recognition and Decentralised Procedures (CMDh). Report from the CMDh meeting held on 22–23 February 2022 (EMA/CMDh/78976/2022), 2 March 2022. Accessed December 07, 2023 at: https://www.ema.europa.eu/en/documents/agenda/agenda-chmp-meeting-21-24-february-2022_en.pdf
• 91 EMA (European Medicines Agency). Committee on Herbal Medicinal Products (HMPC). European Union herbal monograph on Foeniculum vulgare Miller subsp. vulgare var. vulgare, fructus Draft – Revision July 2022 (EMEA/HMPC/372841/2016). Accessed December 07, 2023 at: https://www.ema.europa.eu/en/medicines/herbal/foeniculi-amari-fructus
• 92 EMA (European Medicines Agency). Committee on Herbal Medicinal Products (HMPC) European Union herbal monograph on Foeniculum vulgare Miller subsp. vulgare var. dulce (Mill.) Batt. & Trab., fructus Draft – Revision July 2022EMEA/HMPC/372839/2016. Accessed December 07, 2023 at: https://www.ema.europa.eu/en/documents/herbal-monograph/european-union-herbal-monograph-foeniculum-vulgare-miller-subsp-vulgare-var-dulce-mill-batt-trab-fructus-revision-1_en.pdf
• 93 EMA (European Medicines Agency). Committee on Herbal Medicinal Products (HMPC) Public statement on Foeniculum vulgare Miller subsp. vulgare var. vulgare, aetheroleum Draft – EMA/HMPC/522456/2021. Accessed December 07, 2023 at: https://www.ema.europa.eu/en/medicines/herbal/foeniculi-amari-fructus-aetheroleum
• 94 Pank F, Schneider E, Krüger H. Possibilities and limitations of estragole content reduction of fennel (Foeniculum vulgare Mill.) and its preparations. Zeitschr Arzn Gewürzpfl 2003; 8: 165-172
• 95 EC (European Commission). Directorate-General for Health and Food Safety, Food and Feed Safety, Innovation E1 – Food information and composition ref. Ares (2022)5233398-19/07/2022. Accessed December 07, 2023 at: https://ec.europa.eu/info/funding-tenders/opportunities/docs/2021-2027/horizon/wp-call/2021-2022/wp-10-european-innovation-ecosystems_horizon-2021-2022_en.pdf
• 96 Smith RL, Adams TB, Doull J, Feron VJ, Goodman JI, Marnett LJ, Portoghese PS, Waddell WJ, Wagner BM, Rogers AE, Caldwell J, Sipes IG. Safety assessment of allylalkoxybenzene derivatives used as flavouring substances – methyl eugenol and estragole. Food Chem Toxicol 2002; 40: 851-870
• 97 Anthony A, Caldwell J, Hutt AJ, Smith RL. Metabolism of estragole in rat and mouse and influence of dose size on excretion of the proximate carcinogen 1′-hydroxyestragole. Food Chem Toxicol 1987; 25: 799-806
• 98 Guenthner TM, Luo G. Investigation of the role of the 2′,3′-epoxidation pathway in the bioactivation and genotoxicity of dietary allylbenzene analogs. Toxicology 2001; 160: 47-58
• 99 Jeurissen SM, Punt A, Boersma MG, Bogaards JJP, Yiannis CF, Schilter B, van Bladeren PJ, Cnubben NHP, Rietjens IMCM. Human cytochrome P450 enzyme specificity for the bioactivation of estragole and related alkenylbenzenes. Chem Res Toxicol 2007; 20: 798-806
• 100 Herrmann K, Engst W, Appel KE, Monien BH, Glatt H. Identification of human and murine sulfotransferases able to activate hydroxylated metabolites of methyleugenol to mutagens in Salmonella typhimurium and detection of associated DNA adducts using UPLC-MS/MS methods. Mutagenesis 2012; 27: 453-462
• 101 Suzuki Y, Umemura T, Ishii Y, Hibi D, Inoue T, Jin M, Sakai H, Kodama Y, Nohmi T, Yanai T, Nishikawa A, Ogawa K. Possible involvement of sulfotransferase 1A1 in estragole-induced DNA modification and carcinogenesis in the livers of female mice. Mutat Res 2012; 749: 23-28
• 102 Ishii Y, Suzuki Y, Hibi D, Jin M, Fukuhara K, Umemura T, Nishikawa A. Detection and quantification of specific DNA adducts by liquid chromatography-tandem mass spectrometry in the livers of rats given estragole at the carcinogenic dose. Chem Res Toxicol 2011; 24: 532-541
• 103 Punt A, Delatour T, Scholz G, Schilter B, van Bladeren PJ, Rietjens IMCM. Tandem mass spectrometry analysis of N2-(trans-Isoestragol-3′-yl)-2′-deoxyguanosine as a strategy to study species differences in sulfotransferase conversion of the proximate carcinogen 1′-hydroxyestragole. Chem Res Toxicol 2007; 20: 991-998
• 104 Schulte-Hubbert R, Kupper JH, Thomas AD, Schrenk D. Estragole: DNA adduct formation in primary rat hepatocytes and genotoxic potential in HepG2-CYP1A2 cells. Toxicology 2020; 444: 152566
• 105 Yang X, Feng Y, Zhang Z, Wang H, Li W, Wang DO, Peng Y, Zheng J. In vitro and in vivo evidence for RNA adduction resulting from metabolic activation of methyleugenol. J Agric Food Chem 2020; 68: 15134-15141
• 106 Carlsson MJ, Vollmer AS, Demuth P, Heylmann D, Reich D, Quarz C, Rasenberger B, Nikolova T, Hofmann TG, Christmann M, Fuhlbrueck JA, Stegmüller S, Richling E, Cartus AT, Fahrer J. P53 triggers mitochondrial apoptosis following DNA damage-dependent replication stress by the hepatotoxin methyleugenol. Cell Death Dis 2022; 13: 1009
• 107 Thomas AD, Fahrer J, Johnson GE, Kaina B. Theoretical considerations for thresholds in chemical carcinogenesis. Mutat Res Rev Mutat Res 2015; 765: 56-67
• 108 Doak SH, Jenkins GJ, Johnson GE, Quick E, Parry EM, Parry JM. Mechanistic influences for mutation induction curves after exposure to DNA-reactive carcinogens. Cancer Res 2007; 67: 3904-3911
• 109 Fahrer J, Frisch J, Nagel G, Kraus A, Dörsam B, Thomas AD, Reißig S, Waisman A, Kaina B. DNA repair by MGMT, but not AAG, causes a threshold in alkylation-induced colorectal carcinogenesis. Carcinogenesis 2015; 36: 1235-1244
• 110 Kraus A, McKeague M, Seiwert N, Nagel G, Geisen SM, Ziegler N, Trantakis IA, Kaina B, Thomas AD, Sturla SJ, Fahrer J. Immunological and mass spectrometry-based approaches to determine thresholds of the mutagenic DNA adduct O(6)-methylguanine in vivo. Arch Toxicol 2019; 93: 559-572
• 111 Thomas AD, Jenkins GJ, Kaina B, Bodger OG, Tomaszowski K, Lewis PD, Doak SH. Influence of DNA repair on nonlinear dose-responses for mutation. Toxicol Sci 2013; 132: 87-95

Publication History

Received: 17 September 2023

Accepted after revision: 23 November 2023

Article published online:
10 January 2024