2.1.1. Toxicokinetics (TK)In the last few years, an increasing number of studies have been assessing the human exposure to mycotoxins in different countries using exposure biomarkers. DON, in particular, has been studied frequently because it is mainly excreted in urine: the parent compound and metabolites account for approximately 70% of the total dietary intake [31,32,33,34,35].During the HBM4EU project, a literature search was performed on the TK of DON and FB1–4. Additional searches from other sources than scientific journals (e.g., EFSA reports and opinions) were also performed. Although excretion of DON and its main metabolites in humans has received ample attention in HBM studies, the TK in humans after single or multiple dosing has not been studied frequently. Vidal et al. carried out a human intervention study to unravel the urinary excretion profile and metabolism of DON and its modified form DON-3G [34]. Twenty volunteers were restricted to consuming cereals and cereal-based foods for 4 days. At day 3, a single bolus of 1 μg/kg body weight of DON and a single bolus of 1 μg/kg body weight of DON-3G, after a washing-out period of two months, was administered, and 24 h urine collection was performed. The urine was analysed for DON, DON-3G, 3-ADON, 15-ADON, deepoxy-deoxynivalenol (DOM-1), deoxynivalenol-3-glucuronide (DON-3-GlcA) and deoxynivalenol-15-glucuronide (DON-15-GlcA). The urinary biomarker analysis revealed that DON and DON-3G were rapidly absorbed, distributed, metabolized and excreted. Sixty-four percent of the administered DON and fifty-eight percent of DON-3G was recovered in the urine collected within 24 h. DON-15-GlcA was the most prominent urinary biomarker, followed by free DON and DON-3-GlcA.In a follow-up manuscript [35], the authors developed biokinetic models for DON and DON-3G to determine: (1) the preferred (set of) urinary biomarker(s), (2) the preferred urinary collection period, and (3) a method to estimate the dietary exposure to these mycotoxins. The biokinetic models were based on three physiological compartments (gastrointestinal tract, liver and kidneys) and a known dietary exposure to these mycotoxins (i.e., the single DON or DON-3G bolus). This was used to estimate a reversed dosimetry factor (RDF). The main metabolic pathway for DON elimination is via glucuronidation, with DON-15-GlcA being the major metabolite [34,35,36,37,38]. DON and DON-3G are excreted relatively rapidly: within 12 h, 95% of the total elimination products are excreted via urine.Van den Brand et al. [39] studied the excretion of DON after multiple daily dosing. These authors have assessed the relation between dietary DON intake and the excretion of its major metabolite DON-15-GlcA through time, in an everyday situation. For 49 volunteers from the EuroMix biomonitoring study, the intake of DON from each meal was calculated and the excretion of DON and its metabolites were analysed for each urine void collected separately throughout a 24 h period. The relation between DON and DON-15-GlcA was analysed with a statistical model to assess the residence time and the excreted fraction of ingested DON as DON-15-GlcA (fabs_excr). The estimated time in which 97.5% of the ingested DON was excreted as DON-15-GlcA was 12.1 h and the elimination half-life was 4.0 h. Based on the estimated fabs_excr, the mean RDF of DON-15-GlcA was 2.3. This RDF is comparable to the RDF reported in Mengelers et al. [35], and these can be used to calculate the amount of total DON intake in an everyday situation, based on the excreted amount of DON-15-GlcA. It was also shown that urine samples collected over 24 h was the optimal design to study DON exposure using HBM. So far, no physiologically based toxicokinetic models have been developed for DON.

Little is known on the renal excretion of FB1–4 in humans, presumably because animal studies have shown that FB1 is poorly absorbed from the gastrointestinal tract, rapidly cleared from the blood by the biliary route and preferentially excreted with the faeces. It is generally assumed that the metabolism and excretion of FB2, FB3 and FB4 are similar to that of FB1. The few data on the excretion of FB1 in humans consuming fumonisin-contaminated maize have suggested that the TK of FB1 in humans is similar to other mammalian species.

No TK models have been developed for FB1–4 in humans. Limited information is available regarding the TK of fumonisins in animals and it is mainly related with FB1. Previous studies have concluded that FB1 is poorly absorbed after oral ingestion in farm animals (e.g., swine, cow, laying hen) and experimental animals (rat, mouse, monkey) [29]. The bioavailable amount (less than 4% of the dose) is rapidly distributed to all organs and eliminated by biliary excretion without biotransformation. Faecal excretion vastly predominates over urinary excretion. Small amounts of partly hydrolysed and fully hydrolysed FB1 were detected as metabolites in faeces and are believed to be generated by the colonic microbiome. The EFSA reviewed in its scientific opinion the available data regarding the TK of FB1 [29,40]. According to the EFSA’s opinion, the vast majority of studies on fumonisins have been conducted with FB1 or with a natural mixture of fumonisins obtained from fungal cultures, which contained predominantly FB1 and smaller amounts of FB2 and FB3. No studies have been identified on the TK of FB3 and FB4, and only limited data have been identified on the modified forms HFB1 (hydrolysed FB1), pHFB1 (partially hydrolysed FB1) and NDF-FB1 (N-(1-deoxy-D-fructos-1-yl)-FB1) and no data on NCM-FB1 (N-(carboxymethyl)-FB1), although the latter compound is relevant as it was also detected in food samples. In general, the available studies considered that relating the FB1 concentration in urine to the dietary intake of FB of individual subjects is complicated due to interindividual variability and the rapidity of its clearance [41]. 2.1.2. Main Health Effects Identified for DON and FB1

Exposure to DON or FB1 has been shown to cause various adverse effects in in vitro and in vivo studies.

DON is suspected to be toxic for reproduction and it is able to cross the human placenta [42]. DON exposure affects H295R cell viability, steroidogenesis and gene expression indicating their potential as endocrine disruptors [27]. DON (and other trichothecenes) is immunotoxic, acting as a potent inhibitor of protein synthesis, stimulating the pro-inflammatory response and leading to oxidative stress generation [43].FB1 acts by inhibiting ceramide synthases (CerS), key enzymes in sphingolipid metabolism. Besides being possibly carcinogenic to humans [30], in vivo studies have shown that the repeated exposure to this toxin causes liver and kidney toxicity [29] and may lead to liver and kidney tumorigenesis [30]. FB1 is clastogenic to mammalian cells [29] and also acts as a tumor promoter. Based on the results of animal studies, FB1 has been considered as a potential immunotoxic substance [44].There are, however, no epidemiological studies that can provide unequivocal evidence of the chronic effects of these mycotoxins in humans. Only the acute effect upon DON exposure is well established in humans. With respect to FB1, one epidemiological study suggested an association between maternal FB1 exposure and neural tube defect (NTD) in the fetus [45]. In addition, Marasas et al. noted that several studies reported high occurrence of FB1 in certain areas where high frequencies of NTD in newborns were also observed [46]. This can be considered as circumstantial evidence for an association between FB1 exposure and NTD. Additionally, various animal studies support a link between FB1 and NTD [47,48,49,50,51,52,53,54].Since a solid association between FB1 exposure and NTD could not be established in humans, the AOP framework was used within the HBM4EU to collect and organize in vitro and in vivo studies to support the circumstantial evidence on possible adverse effects of FB1 in humans [55,56]. 2.1.3. Development of AOPs for FB1 Based on Mechanistic KnowledgeAn AOP describes the main events leading from a perturbation at the molecular level (the molecular initiating event, MIE) to an adverse effect on the organism or population (the adverse outcome, AO). The AOP framework helps to organize the data and evaluate the evidence that an event causes the next one. This curated mechanistic knowledge is accessible to users (e.g., risk assessors) through the AOP knowledge base (the AOP-KB, https://aopkb.oecd.org/index.html, accessed on 22 July 2022) and wiki (the AOP Wiki, https://aopwiki.org/, accessed on 22 July 2022). The AOP describing the biological mechanism that may underlie FB1-induced NTD in the developing embryo [55] (ID 449 in the AOP Wiki) was based on the mode of action described in the EFSA’s scientific opinion [29] and by Marasas et al. [46].It proposes that FB1 triggers its effects (molecular initiating event, MIE) by inhibiting CerS, a key enzyme in sphingolipid metabolism (Figure 2). Indeed, FB1 is a structural analog of sphinganine (Sa) and sphingosine (So), the substrates of CerS, and a well-known inhibitor of CerS [57,58]. The drafted AOP describes two possible chains of events leading from CerS inhibition to NTD. The first route involves a decrease in folate uptake, which is known to be associated with the frequency of NTD. CerS inhibition may impact folate uptake through a decrease in complex sphingolipids (e.g., gangliosides), which are important constituents of membrane microdomains to which the folate receptor is presumably anchored [49,59,60,61]. The second route involves the inhibition of histone deacetylase (HDAC), which may induce NTD, as depicted in an existing AOP (under development) (ID 275 in the AOP-wiki, https://aopwiki.org/aops/275, accessed on 22 July 2022). In this scenario, HDAC activity is inhibited by phosphorylated forms of Sa and/or So, which are expected to increase in response to CerS inhibition [48,62,63].

Therefore, by describing biological mechanisms for FB1-induced NTD through inhibition of CerS, the proposed AOP reinforces the observations from limited human studies and provides a rationale for a causal relationship between exposure and health. It also highlights some uncertainties and gaps in the existing knowledge. For instance, it remains to be established how the first steps (key events) lead to a decrease in folate uptake, and the applicability of the chain of events in humans, given that the mechanism is largely based on in vitro and animal studies. Finally, it will be highly important to define the threshold levels of FB1 needed to trigger the MIE and compare them with predicted internal FB1 concentration.

Based on the mode of action proposed in the most recent EFSA scientific report, a putative AOP can be drafted for the DON-induced reduction of body weight gain [26,55]. The proposed MIE is the binding to ribosomes that would activate mitogen-activated protein kinases (MAPK), leading to the effects on body weight through two possible mechanisms. The first involves an inflammatory response in the intestine while the second involves the secretion of gut satiety hormones. These mechanisms find solid support from several in vitro and animal studies, but evidence for this association between DON exposure and reduced body weight gain in human is lacking. 2.1.4. Effect Biomarkers in HBM StudiesIn the scope of the HBM4EU Project, a literature search was carried out to find the most frequently used effect biomarkers in epidemiological studies addressing DON or FB1 exposure through HBM [64,65]. Even though the number of HBM studies focused on mycotoxin exposure has been increasing in the last decade, only very few of them have included the analysis of effect biomarkers. Several studies have suggested that the increase of the Sa/So ratio in biological fluids can be used as a sensitive biomarker of fumonisin exposure and early biological effects [41,66,67]. The validity of the Sa/So ratio as a biomarker in humans remains, however, uncertain [67,68]. This is partly because Sa and So occur and vary naturally in human blood [68,69]. Furthermore, there is no human reference value for physiologically normal levels of these sphingoid bases or the Sa/So ratio. In this context, the proposed AOP provides some mechanistic support for the Sa/So ratio as a biomarker of an early effect. However, although the increase in Sa/So ratio is an expected consequence of the MIE, it is not a key event of the AOP per se. Based on the AOP that was developed for FB1, other effect biomarkers may be proposed. As FB1 is recognized as an inhibitor of HDAC, a central regulator of gene expression, a more open chromatin structure, resulting in deregulated gene expression is expected following exposure to this mycotoxin. Therefore, in vitro assays that target chromatin structure or downstream gene expression can be used to start the development of novel effect biomarkers.

Concerning DON exposure, no effect biomarkers were found that could be associated with its adverse health effects.

2.1.5. Derivation of a HBM-GV for DONWithin the HBM4EU, a HBM-GV of 23 μg total DON/L (95% CI 5–33 μg total DON/L) was derived for total DON concentrations in 24 h urine samples [70]. Depending on the analytical approach taken to analyse the urine, an HBM-GV of 20 μg DON-15-GlcA/L (95% CI 7–39 μg DON-15-GlcA/L) can also be applied for this main (phase II) metabolite of DON [70]. As more than 90% of DON is excreted approximately 12 h after ingestion, almost all ingested DON is captured in a 24 h urine sample (provided that it includes the first morning urine sample at the end of the collection period) [35,39].The HBM-GV for DON was derived for 24 h urine samples and should not be applied to morning urine or other spot urine samples, as the elimination half-life of DON is too short (approximately 3 to 4 h). The variation in spot urine samples will consequently be too large (larger than the dosing interval calculated for a 24 h urine sample). This, of course, has practical consequences, as many studies (including the HBM4EU Aligned Studies on DON) collected spot urine samples rather than 24 h urine samples [71,72]. Although there are studies that report a good correlation between morning urine and 24 h urine samples [73], a fraction to convert a morning urine sample to a 24 h urine sample cannot be used. The uncertainty around this fraction would be too large considering the short excretion half-life of DON, the variation of DON intake throughout the day and the lack of information regarding the urine voids/discharges of the volunteers between their last meal and first morning urine collection. In addition, the HBM-GV was derived by using an average, body-weight-adjusted, urinary flow rate based on the 24 h urine volumes of the 20 volunteers in a human intervention study [35]. This flow rate can differ between populations, which will affect the derived HBM-GV. Considering that the study population of Mengelers et al. [35] was small, the average flow rate may be prone to fluctuations. This is a general observation that needs to be kept in mind when deriving a HBM-GV from a HBGV that was based on an animal study, only using ‘external’ doses. 2.1.6. Responses to Policy Questions on Hazard Assessment

Concerning the policy question defined under the HBM4EU, “Are there toxicokinetics data for the target mycotoxins and which are their limitations?”, the answer is affirmative for DON, since a dedicated model was developed under the HBM4EU, but not for FB1. Future efforts should be made to increase the knowledge on FB1 toxicokinetics and, consequently, contribute to a better human risk assessment. Currently, the toxicokinetics of DON after oral intake appear to be better characterized. Nevertheless, occupational exposure to DON can occur through inhalation of, e.g., contaminated flour dust, and the absorption characteristics for this route of exposure are still unknown. Physiologically based toxicokinetic models, including absorption by inhalation and oral absorption could contribute to the risk assessment of occupational exposure.

In response to the question “Which are the key events that determine the chronic health effects of the target mycotoxins?”, an AOP was provided, in which several key events might determine FB1-induced neural tube defects. However, data on long-term effects from low-dose continuous exposure are lacking and human studies are insufficient. Moreover, associations between key events/potential effect biomarkers and the health effects are not established. In fact, there are difficulties in reliably establishing exposure to FB1 in humans, since some key events are technically challenging, and causality for chronic effects is difficult to establish with human data.

Concerning the question “Which are the most frequent AOP-based effect biomarkers for the prioritized mycotoxins?” and after a literature review, the most specific effect biomarker identified was related to the inhibition of CerS: Sa and So levels in blood or urine. However, there are insufficient data available on the effect biomarkers for FB1 and no studies for DON. For example, omics-based methods, such as transcriptomics, can be used to identify a number of differentially expressed genes involved in previously identified KE that can be candidate effect biomarkers. The value of those candidates can start to be evaluated through in vitro functional assays and their sensitivity and reliability can then be assessed in epidemiological studies.

Finally, concerning the question “Is it possible to set a HBM-GV for the target mycotoxins”: indeed, a HBM-GV was derived for DON, but it is only applicable for 24 h urine samples, which increases the uncertainty associated with its use for spot urine samples.