Mycotoxins and pyrrolizidine alkaloids in herbal dietary supplements

ABSTRACT The market demand for herbal dietary supplements is rapidly growing and such products are becoming more common and accessible to consumers. However, the knowledge about their safety remains incomplete. Herbal dietary supplements are one of the food groups that can contribute significantly to human health concerns arising from chronic exposure to pyrrolizidine alkaloids and mycotoxins. This study aimed to simultaneously determine 79 natural contaminants, including mycotoxins, as well as pyrrolizidine and tropane alkaloids in herbal dietary supplements in one analytical run. Exposure assessment and human health risks were assessed for all compounds included in this study. The total concentration of naturally occurring contaminants in herbal dietary supplements reached 5.3 mg kg−1 and the most frequently detected mycotoxins were tentoxin and alternariol monomethyl ether. The latter was detected with the highest frequency, reaching concentrations up to 2.5 mg kg−1. The obtained results indicate a potential risk to public health related to herbal dietary supplement consumption.


Introduction
The collection and use of various plants for medicinal purposes has been a traditional part of human culture that is still important today.In recent years, standardised herbal medicine have entered pharmacies, supermarkets and online stores as herbal dietary supplements (HDS).The global vitamin and food supplement market is growing due to increasing consumer interest in personal health and well-being (Pallarés et al. 2022).In addition, COVID-19 pandemic positively impacted the demand for vitamins and dietary supplements on a global scale (Kalkuz and Göktaş 2023).In general, dietary supplements contain a wide range of nutrients and other components, including, but not limited, to vitamins, minerals, amino acids, essential fatty acids, fibres, antioxidants, various plant materials and excipients (Baran 2014).They may be marketed in the form of tablets or capsules containing a mixture of homogenised plant parts (leaves, roots, flowers) or dried ethanolic extracts, as well as liquids, e.g.syrups, infusions or oils.For various reasons, the final product may contain certain contaminants, such as pesticides, heavy metals, moulds, pyrrolizidine alkaloids.Although the quality of the raw materials is of great importance, the storage and production conditions also affect the safety of finished formulation.Humid and warm storage conditions favour the proliferation of various moulds, leading to mycotoxin contamination (Pilarska et al. 2022).
Several previous studies have reviewed the occurrence of mycotoxins in HDS and their ingredients.Thus, Silybum marianum (L.) Gaertn, commonly known as Milk thistle, along with various herbal teas and botanicals, have been repeatedly shown to contain relatively high concentrations of mycotoxins (Pickova et al. 2020;Pallarés et al. 2022), mainly from Aspergillus, Alternaria, Fusarium and Penicillium fungi (Zhang et al. 2018).The current Regulation (EU) 2023/915 of the European Commission (2023) sets maximum levels of mycotoxins in various foodstuffs, but for HDS or their ingredients only the content of citrinin and PAs is regulated.As a result of that, many studies focus only on the prevalence of regulated mycotoxins like aflatoxins, ochratoxin A, trichothecenes, zearalenone and fumonisins.However, emerging mycotoxins such as the enniatin group and Alternaria mycotoxins (Figure 1) can occur in HDS and constitute a significant part of the total toxin burden (Veprikova et al. 2015;Pickova et al. 2020;Pallarés et al. 2022).
HDS is one of the food groups that can contribute significantly to the human health risks associated with chronic cumulative exposure to pyrrolizidine alkaloids (PAs), which are plant-produced toxic compounds that can cause liver damage in humans (KnutsenHK et al. 2017).The diversity in structure within the necin base and its derivative forms can be extensive (Figure 2).The pyrrolizidine core -two saturated five-membered rings with a nitrogen atom between them -occasionally exhibits a double bond in the 1,2 position, often leading to increased toxicity (Moreira et al. 2018).Not all plants used in herbal medicine initially contain PAs, but the final product can be cross-contaminated with PAs from other plants or weeds during harvesting (Schrenk et al. 2020).Recently, the maximum level of PAs in food supplements containing botanical preparations has been set at 400 µg kg −1 by the European Commission (European Commission 2023).This limit is defined as the sum of 35 PAs, including 21 compounds and 14 isomers, that co-elute during chromatographic separation (European Commission 2023).Therefore, low LOQ levels of <5 µg kg −1 must be achieved, representing a major challenge for the analysis of such complex and diverse samples as HDS.
The tropane alkaloids atropine and scopolamine occur naturally in the plant family of Solanaceae.These compounds are toxic and can be present in HDS as contaminants in the same way as PAs (Figure 3).Atropine and scopolamine have been reported in herbal teas (Shimshoni et al. 2015) and herbal infusions (Mateus et al. 2023).Regulation (EU) 2023/915 (European Commission 2023) also imposes maximum levels of tropane alkaloids in various food products.The limit for the sum of atropine and scopolamine in herbal infusions (dried product) has been set at 25 µg kg −1 , except for herbal infusions made exclusively of anise seeds (50 µg kg −1 ).
From the legal point of view, food supplements are considered as food in the EU (European Commission 2023).These products are available to every consumer in grocery stores, pharmacies and even online.Most HDS are offered with various health-enhancing claims, such as improving the nervous system, digestive tract health, liver function, cardiovascular health and other benefits.This indirectly allows the consumer to self-medication without consulting a doctor.An additional concern is that the recommended duration of use for such herbal dietary supplements may be up to several months, as amounting to a long-term exposure.The quality control mechanisms for herbal medicines have become a major concern of health authorities and the public.This highlights the need to develop new, sensitive, broad-spectrum methods for monitoring naturally occurring contaminants in various herbal dietary supplements.
Several relevant multi-analyte analytical procedures have been reported (Sarkar et al. 2022;Liao et al. 2023;Rizzo et al. 2023), nevertheless, simultaneous determination of various contaminant groups has been less  frequently considered (Romero-González et al. 2011;Lacina et al. 2012;Dzuman et al. 2015).QuEChERS (Quick, Easy, Cheap, Effective, Rugged and Safe) extraction is a well-known and widely adopted method for sample preparation in food safety control (Santos et al. 2022;Jansons et al. 2022;Makni et al. 2023).It has been implemented in mycotoxin research (Dong et al. 2018;Reinholds et al. 2020) and many studies have confirmed its suitability for PA determination (Jansons et al. 2022;León et al. 2022;Casado et al. 2022).Similarly, chromatographic separation on reversed-phase columns and detection can be applied for PA determination, using either Orbitrap or triple quadrupole mass spectrometers (Dzuman et al. 2015;Augustin Mihalache et al. 2023;Zan et al. 2023).This study describes an analytical method for the simultaneous determination of 79 natural contaminants, including mycotoxins, pyrrolizidine alkaloids and tropane alkaloids in herbal dietary supplements from the Latvian retail markets and online stores.With the obtained data a preliminary safety assessment of some of these was carried out, to demonstrate that its consumption substantially contributes to total mycotoxin exposure.
Individual stock solutions for each compound at concentrations of 250-1000 µg mL −1 were prepared in methanol for pyrrolizidine alkaloids and in acetonitrile for mycotoxins.Working standard solutions for each compound group were prepared at 0.020-10.0µg mL −1 in acetonitrile.All stock solutions were stored at −20°C, while working standard solutions were stored at + 4°C.

Sampling
Samples of herbal dietary supplements were collected from Latvian retail markets and online stores.A precise description of the samples regarding the composition and dosages of the preparations has been summarised in the Supporting information (table S1).All 47 collected types of HDS were in the form of tablets (16 samples) or dry powders (31 samples), including capsules with powder filling (24 samples).Most of them contained dry plant extracts (up to 11 botanicals) with vitamins, bioactive compounds and excipients.The sample origins included mostly countries of the European region, but also India, USA, Uzbekistan and Israel.The majority of collected types of HDS were intended to improve the health of various organs (Figure 4), such as the functioning of the digestive tract, improving liver health, strengthening the nervous system.

Sample preparation
Prior to analysis, samples were stored according to the instructions on the label.Before sample homogenisation, 6 tablets were weighed and the average weight per tablet was calculated.The tablets were ground in a spice and coffee grinder (type MC741CG with 70 g capacity from supplier Krinona, Kaunas, Lithuania) until homogenous powder.In case of capsules, only the powder inside was used for analysis.Therefore, 6 capsules from each sample were carefully opened and their contents were weighed to obtain an average dosage weight.Information about dosage weight is summarised in Supporting information Table S2 and also used for the Estimated Daily Intake (EDI) calculations.
The samples were extracted according to a QuEChERS-based method, similarly as published before (Jansons et al. 2022).In brief, a 1.00 g sample was weighed into a 50 mL polypropylene (PP) centrifuge tube (Sarstedt AG & Ko., Nümbrecht, Germany).Aqueous 0.2% formic acid (10.0 mL) was added to the dry sample.The samples were shaken for 30 minutes, as suggested elsewhere (Veprikova et al. 2015;Zhang et al. 2018).After that, 10.0 mL of acetonitrile was added and shaken for 10 min.QuEChERS salts (4.0 g magnesium sulphate, 1.0 g sodium chloride, 1.0 g tribasic sodium citrate dihydrate and 0.5 g dibasic sodium citrate sesquihydrate) were added to the tubes with extracts and shaken for 10 minutes.After centrifugation (10 min at 4700 rpm, Multifuge X3R, Thermo Scientific, Waltham, MA, USA), the upper acetonitrile layer was transferred to a 15 mL PP centrifuge tube.In order to separate non-polar compounds, the sample tubes were frozen at −70°C (Heto-Holten, Allerød, Denmark) and centrifuged at 4700 rpm for 15 min at 20°C.Aliquots (3 mL) were transferred to new 15 mL PP centrifuge tubes for evaporation under a gentle nitrogen stream at 45°C in a water bath.The dry residues were dissolved in the HPLC injection solution containing 25% acetonitrile and 1% formic acid in water.The final extracts were filtered through Ultrafree-MC centrifugal filters (0.22 µm pore size, hydrophilic PVDF) and transferred to chromatography vials.Instrumental analysis was performed immediately.

Quantification
All mycotoxins were identified as individual compounds, except 3-and 15-acetyldeoxynivalenol, which were identified and quantified as a sum of both compounds, due to lacking chromatographic separation, which is a wellknown phenomenon (De Girolamo et al. 2020).Likewise, most pyrrolizidine alkaloids could be identified individually, except the pairs of indicine and lycopsamine, indicine N-oxide and intermedine N-oxide, integerrimine and senecivernine and usaramine and retrosine, which were therefore identified and quantified as unresolved pairs.
At first, a screening analysis was performed to assess the approximate levels of contaminants in each sample.For quantification, samples were prepared by standard addition prior to sample preparation using 3-level calibration, according to the concentration determined in the screening step.

Separation and detection
An UltiMate™ 3000 system (Thermo Fisher Scientific, Waltham, MA, USA) equipped with a Phenomenex Luna Omega C 18 reversed-phase analytical column (100 × 2.1 mm, 1.6 µm), coupled to a TSQ Quantiva detector was used for the analysis.Dual gradient elution was carried out using 0.1% formic acid and 0.5 mM ammonium acetate in water (eluent A) and acetonitrile (eluent B) according to the following gradient programme: equilibration time 6.0 min at 1% of B; linear change from 1% to 35% B in 20 min and from 35% up to 99% B in 6 min; equilibration at 99% of B for 9 min.The flow rate was 0.30 mL min −1 .The autosampler was maintained at 15°C and the column temperature was 40°C.The sample injection volume was 5 µL.Ion monitoring was conducted in both positive and negative ion mode and the mass analysis was performed in the selected reaction monitoring (SRM) mode.The SRM method parameters and retention times are summarised in the Supporting information, Table S3.The following instrumental settings for the mass spectrometer were used: spray voltage 3.5 kV (positive ion mode), 2.5 kV (negative ion mode), vaporiser temperature 350°C, ion transfer temperature 300°C, sheath gas 55 arbitrary units (arb), auxiliary gas 25 arb and sweep gas 5 arb.Data processing was performed with ThermoTraceFinder EFS LC software, also from Thermo Fisher Scientific.

Estimated daily intake and hazard quotient
The Estimated Daily Intake (EDI) levels of mycotoxins and PA were calculated and compared with the Tolerable Daily Intake (TDI) or other safety threshold values obtained from the available toxicology data in order to assess the food safety risks associated with the identified contamination.
The EDI levels were obtained by including the number of daily doses recommended by the manufacturer, the weight of one dose and the determined concentration of contaminants in the calculations, as shown in Equation 1.The recommended dosage information for all tested HDS products is summarised in Table S2.EDIs were expressed in ng/day and an average adult body weight of 70 kg was used in the calculations for exposure assessment (EFSA Scientific Committee 2012), resulting in the formula EDI = [(γ x N HDS x m HDS )/70] x 1000, where γ is the concentration of natural contaminants in a sample (µg kg −1 ); N HDS is the daily HDS dosage as recommended by the manufacturer, tablets/capsules per day; and m HDS is the weight of 1 HDS dose (kg).In all calculations, a lower bound approach was used, where a value of 0 µg kg −1 was assigned to samples where mycotoxin and PA levels were below LOQ.
Hazard quotient (HQ) was used to characterise the risk associated with the mycotoxin content in HDS.HQ values were calculated by HQ = [EDI/(TCC or BMDL)] x 100%, where TCC is the Threshold of Toxicological Concern (ng/kg bw/day) and BMDL is the benchmark dose (lower confidence limit; ng/kg bw/day).The dietary risk was considered to be acceptable when HQ < 100%, otherwise (HQ > 100%) the contaminant was deemed to present a potential threat to public health.

Method validation
For validation of the method performance, sample No.29 was selected as a blank since it contained low native PA and mycotoxin contamination.Five-point calibration in the range 1-50 µg kg −1 for PA and 12.5-500 µg kg −1 for mycotoxins, except for deoxynivalenol (250-2000 µg kg −1 ) and the sum of 3-and 15-acetyldeoxynivalenol (500-4000 µg kg −1 ) were prepared.In order to evaluate the linear working range, procedural calibration curves were constructed for calculating the linear regression coefficients (R 2 ).For all studied analytes the R 2 values were in range 0.952 to 0.990.
The repeatability of the method was expressed as the relative standard deviation (RSD) of six replicates of spiked samples, Table 1.The recoveries ranged from 86 to 119% and the relative standard deviation for almost all (96%) compounds was ≤ 20%, as recommended by the Commission Decision for regulated mycotoxins (European Commission 2006).The signal -to -noise (S/N) approach was used to estimate LOD and LOQ values.The LOQ values were estimated as the lowest procedural calibration level that provided S/N ≥ 10 and the criterion for LOD was S/N ≥ 3. The LOQs ranged from 0.25 to 500 µg kg −1 , similar to previously reported LOQs achieved for PA (Rizzo et al. 2023) and for mycotoxins (Veprikova et al. 2015) in food supplements.
For evaluating the matrix effect (ME), three-point standard calibrations in injection solution and matrixmatched calibration curves were prepared.The ME values were calculated to evaluate a possible signal suppression or enhancement (SSE).SSE was calculated by SSE = [(Slope matrix/Slope solvent) − 1] x 100%.A positive result indicates ion enhancement and a negative result ion suppression.The obtained SSE values were in the range −93 to 31%, whereas the median was −44%.The present study confirmed earlier findings about the predominantly observable ion suppression effect in the analysis of plant-based samples (Veprikova et al. 2015;Dzuman et al. 2015).
The extraction efficiency (EE) was studied to ensure the suitability of the extraction method.For this purpose, EE was expressed as the ratio between the average of 3 replicates of the signal area obtained from a spiked sample before preparation and the average signal area of a spiked extract after sample preparation.The average EE for all compounds was 71%, which was in line with previous studies for PAs (Jansons et al. 2022;Rizzo et al. 2023) and mycotoxins (Bessaire et al. 2021), showing that for the quantification of this type of samples and analytes, the standard addition method is most suitable.The obtained validation data are summarised in Table 1.

Mycotoxin and pyrrolizidine alkaloid levels
As reported before, dietary supplements tend to contain a wide range of natural toxins (Veprikova et al. 2015).Likewise, in this study 79% of the samples (37 out of 47) contained detectable levels of mycotoxins or pyrrolizidine alkaloids.All three contaminant groups included in this study, mycotoxins, pyrrolizidine alkaloids and tropane alkaloids, were detected in HDS at total concentrations ranging from <LOQ to 5 mg kg −1 .Figure 5 represents the obtained mycotoxin and pyrrolizidine alkaloid content in positive samples.Tropane alkaloids are not included due to their low concentrations.Different types of HDS according to the claimed health effects were selected for analysis.Although no relationship between the content of mycotoxins or PAs depending on the type of HDS has been previously reported, samples containing milk thistle (most often recommended for improving liver health) were expected to be contaminated with mycotoxins at high levels (Pickova et al. 2020).The results showed that the type of HDS recommended for liver health improvement often contained high concentrations of mycotoxins, but not necessarily the highest.HDS samples recommended for general health improvement contained only a few botanical ingredients (1 to 4), which might explain the significantly lower content of contaminants.
Tentoxin was the most frequently detected mycotoxin (57%), although the mean concentration in positive samples was low (32 µg kg −1 ).Alternariol monomethyl ether and beauvericin were found in 40% of the samples, with mean concentrations of 173 µg kg −1 and 15 µg kg −1 , respectively.The presence of enniatins ranged from 25 to 38% of all samples, with a higher incidence of enniatins B and B1 than enniatins A and A1.Similarly, the maximum concentrations of these mycotoxins followed the  order enniatin B1 > B >> A > A1 (Veprikova et al. 2015).
The mean and even the maximum concentrations of enniatins B1 and B in this study were much lower than reported before (Pallarés et al. 2022).Other Fusarium mycotoxins (T-2, HT-2, zearalenone) were found only in a few samples at low concentrations.Regulated mycotoxins such as aflatoxins were detected in 3 samples, at levels of 0.26-2.46µg kg −1 .Ochratoxin A was found in 2 samples at 6.4 µg kg 1 and 5.2 µg kg −1 and deoxynivalenol in 1 sample (511 µg kg −1 ).The total mycotoxin concentration in 4 samples exceeded 1 mg kg −1 .One HDS sample (No 42.), intended for cardiovascular health improvement, contained 4.9 mg kg −1 of total mycotoxins.The information on the label notified that it contained six plant extracts, including barberry (Berberis aristata DC.), bark dry extract, pomelo fruit extract (Citrus grandis L. Osbeck,), cassia leaf dry extract (Chamaecrista nomame Sieber H.Ohashi), olive fruit dry extract (Olea europaea L.) and milk thistle (Silybum marianum Gaertn.), the latter ingredient having been previously associated with comparable results (Veprikova et al. 2015;Pickova et al. 2020).Interestingly, this sample contained monacolin from fermented red yeast rice (Monascus purpureus) that is commonly used as a lipid-lowering agent in food supplements.Although potentially high levels of citrinin were previously reported in such products, other mycotoxins therein have been rarely found before (Righetti et al. 2021).Likewise, in 4 other samples containing red fermented rice, the mycotoxin content was low (<100 µg kg −1 ), indicating that the source of mycotoxin contamination could be other ingredients.Sample number 33, with the second highest concentration of total mycotoxins (1.4 mg kg −1 ), contained 6 different plant extracts, including milk thistle (Silybum marianum Gaertn.),reishi mushroom dry extract, maral root powder (Rhaponticum carthamoides L.), thyme powder (Thymus serpyllum L.), Melissa leaf powder (Melissa officinalis L.) and camomile (Matricaria chamomilla L.).Taking into account the different compositions of the analysed HDS samples, the obtained results are comparable to those previously reported (Veprikova et al. 2015;Pallarés et al. 2022;Pilarska et al. 2022).

Co-occurrence of mycotoxins
Mycotoxin co-occurrence has been widely reported before in the analysis of herbal products (Zhou et al. 2023).Although the individual mycotoxin content is low in most samples, the toxic effects resulting from the co-occurrence of multiple mycotoxins can be additive and even synergistic, amounting to a potential threat to consumer health (Liang et al. 2015;Stoev 2023).In this study 19% of the samples contained 2 to 5 mycotoxins, 32% contained 6 to 9 mycotoxins and 11% contained 10 or more mycotoxins.In total, 62% of the samples contained more than 1 mycotoxin and the most commonly found were those from Alternaria and Fusarium moulds (Table 2).Occasionally mycophenolic acid (MPA) is found together with Alternaria mycotoxins (Crudo et al. 2019).The two most frequently found mycotoxin combinations were tentoxin + alternariol monomethyl ether and enniatin B + enniatin B1.Four enniatins are commonly found in combinations (Crudo et al. 2019).In this study, correlations of individual enniatins were observed.Although no correlation between MPA and AME has been previously reported, in this study strong correlation was observed between these mycotoxins.

Plant toxins
Tropane alkaloids atropine and scopolamine were found in a few samples, with mean concentrations of 1.8 µg kg −1 (3 samples) and 1.5 µg kg −1 (7 samples), respectively.Ashwagandha powder (No. 21) had the highest atropine content of 4.6 µg kg −1 .Maximum limits for tropane alkaloids have been set by the European Commission (2023) for various foodstuffs, including herbal infusions, but not for food supplements specifically.Tropane alkaloids have been studied most in various cereals, herbal infusions and teas.Reported levels ranged from nanograms per kilo to a few hundred micrograms per kilo (Shimshoni et al. 2015;González-Gómez et al. 2022), depending on the tropane alkaloid producing plant.Samples wherein atropine and scopolamine were detected did not contain plants known to be tropane alkaloid-producing, so their presence might be explained by cross-contamination from weeds or not labelled herbs, similar to the situation with PAs (Schrenk et al. 2020).
The majority of the pyrrolizidine alkaloids (23 of 29) included in the analytical scope were detected in the analysed samples.Heliosupine (26%) and heliotrine (30%) were quantified with the highest incidence, which has also been observed before (Kaltner et al. 2020).Echinatine was determined at the highest concentrations, with 191 µg/kg on average and reaching a maximum of 790 µg kg −1 .Pyrrolizidine alkaloids were found above LOQ in 27 samples (57%).Total PA concentrations in positive samples were in the range of 0.62-1097 µg kg −1 , as in line with previous reports (Mulder et al. 2017;Suparmi et al. 2020;Rizzo et al. 2023).Although no significant pattern was observed in relation to PA and PA N-oxides in the samples, only echimidine, echinatine, and europine were determined more frequently and in higher concentrations than their N-oxides.Attention should be paid to the fact that the distribution of PAs can be extremely diverse, even within Europe.For example, in honey (n = 14) purchased in North Macedonia overall contamination ranged from 89 to 8188 µg kg −1 for total PA/PANOs (15 compounds tested) (Cvetanoska et al. 2024), but in honey purchased in Latvia (n = 40), the highest contamination level was only 74 μg kg −1 (30 compounds tested) (Jansons et al. 2022).
Two samples exceeded the 400 µg kg −1 limit of PAs that has been set by the European Commission for food supplements containing botanical preparations (European Commission 2023).Sample No. 4 contained 1097 µg kg −1 of total PAs, with echinatine contributing the most of it (790 µg kg −1 ).This supplement was manufactured in Latvia and claimed to have health benefits for the nervous and cardiovascular systems.It contained dry extract of hawthorn leaves with flowers, dry extract of valerian roots and dry extract of lemon balm leaves.This product was sold in packages containing 60 capsules and manufacturer's suggestion was to use 2 capsules per day.Sample No. 16 contained 466 µg kg −1 of total PA, mostly summed up with an input of 371 µg kg −1 from heliosupine, heliotrine, intermidine, lasiocarpine, europine and echinatine.According to the label it contained Prickly Caper Root Powder, Common Chicory Seed Powder, Iron Oxide (Mandur Bhasma), Black Evening Primrose Seed Powder, Terminalia Bark Powder, Cassia Seed Powder, Yarrow Powder, Gallic Tamarind Powder and excipients.This product was made in India and was marketed to improve liver health.These ingredients were packed in tablets and manufacturer's suggestion was to use 4 capsules per day.The ingredients in the PA-positive samples were mainly raw plant or dry plant extracts of herbal products well-known to be contaminated with significant levels of PAs (yarrow, camomile, thyme, hawthorn, valerian) (Schulz et al. 2015;Chen et al. 2022).Although no well-known echinatine producing plant names were included in the lists on the labels, high echianitine concentrations in the samples, lead to similar conclusions as drawn by Kaltner et al (Kaltner et al. 2020), that even when no PA/PANO producing plants are used in HDS production, high levels of contamination might occur.
Although other dietary supplements did not exceed the maximum level of total PAs, samples No. 13 and No. 24, converted to a daily dose, showed the highest estimated daily intake of PAs.HDS No.13, intended for weight loss, contained Plantago ovata seed coat powder and glucomannan/Konjac mannan (Amorphophallus konjac) powder, whereas sample No. 24, containing Sand psyllium seed fibres (Plantago psyllium) and Cyamopsis tetragonolobus seeds (endosperm) along with various types of Lactobacillus bacteria, was intended for improving the health of the digestive system.None of these botanicals have previously been reported as PA-contaminated.The total PA content of these samples was not significantly high, 190 and 64 µg kg −1 , respectively, but when the daily dosage is taken into account the EDI of total PAs per day was almost 2 times higher than for samples No. 4 and No. 16, which exceeded the limit of 400 µg kg −1 .The label claimed that both dietary supplements were suitable for children (older than 6 years old) as well, with recommended dosage being only half of the daily amount for adults.
Similar patterns were observed with other powder supplements and mycotoxin content.Since, their daily dosages were higher, the exposure related to mycotoxin and pyrrolizidine alkaloid contamination was higher.That would suggest the necessity not only to focus on ingredients and benefits, but as well on food supplement form, when choosing such products for the long term use.

Exposure and risk assessment
The exposure and risk assessment were calculated for only the most frequently detected mycotoxins.According to EFSA (EFSA on Contaminants in the Food Chain CONTAM 2011), alternariol monomethyl ether is classified as a genotoxic compound and the threshold of toxicological concern (TTC) is set at 2.5 ng/kg bw/day.However, tentoxin has shown negative results in geblic health related to the presence of Alternaria toxins innotoxicity assays, therefore the TTC value for this type of non-genotoxic substance is 1500 ng/kg bw/day (EFSA on Contaminants in the Food Chain CONTAM 2011).
The available data on the toxicokinetics of beauvericin and enniatin groups are limited, as well as the results of genotoxicity assays are ambiguous.In order to obtain some insight into the possible health risks associated with exposure to beauvericin and the sum of enniatins (ENNs), a worst-case scenario was chosen and the same TTC values for compounds containing a structural alert for genotoxicity were used, i.e. 2.5 ng/ kg bw/day for beauvericin and 1500 ng/kg bw/day for the enniatins (EFSA CONTAM Panel EFSA Panel on Contaminants in the Food Chain 2014).
For AME, the obtained EDIs (from positive samples only) were in the range of 2.4-2479 ng/day, resulting in HQs of 1.4-1417%, respectively (Figure 6, Table 3).Four of the 19 positive samples exceeded the TTC value and 5 samples reached 60% or more of the TTC.The samples with the highest EDIs were not similar in composition or purpose of use.Sample No. 42 was a multi-component HDS, including marjoram, sensitive pea, dry extracts of olive and various fruits and red yeast rice, for the improvement of cardiovascular health.Sample No. 40 was ashwagandha powder for general health improvement, sample No. 34 was a multi-component mixture, including cereals, fruits and vegetable fibre, for digestive system health and sample No. 9 was a dry extract of milk thistle for liver health.
The HQ values obtained for beauvericin (BEA) ranged from 0.5 to 123%, exceeding 100% in sample No. 9 only (Table 3).These results are in a good agreement with published studies (Veprikova et al. 2015;Pallarés et al. 2022), where food supplements containing milk thistle revealed the highest mycotoxin content.Regarding the enniatins, the highest EDI value was obtained for sample No. 42, containing 1.04% of the HQ (1500 ng/kg bw).The same TTC value was recommended for the assessment of tentoxin exposure, which resulted in HQ values below 0.18%.
The maximum level of pyrrolizidine alkaloids in food supplements containing botanicals is set at 400 µg kg −1 (European Commission 2023) and the benchmark dose (lower confidence limit) BMDL of 273 µg/ kg bw has been used in risk assessment procedures (KnutsenHK et al. 2017;Jansons et al. 2022).As many as 10.6% of the samples exceeded 400 µg kg −1 (Table 3), almost similar to the previously reported 6% (Rizzo et al. 2023), but the obtained EDI values of the sum of the PAs were very low and contributed less than 0.02% of the BMDL.Detailed information is summarised in Table 3 and Table S2.

Conclusions
This study provided insight into the degree of contamination with mycotoxins, pyrrolizidine alkaloids and tropane alkaloids in various dietary supplements containing at least one plant ingredient.In 79% of the 47 investigated samples of different composition, origin and form the total contamination ranged from <LOQ to 5 mg kg −1 .As the consumption doses for dietary supplements are low (few grams per day), the quantitative daily intake of the contamination is typically negligible.However, the obtained results pointed to potentially significant health hazards in the case of one genotoxic mycotoxin, i.e. alternariol monomethyl ether, which was frequently (40%) found and thus at high concentrations (up to 2.5 mg kg −1 ), consequently the EDI exceeded up to 14 times the threshold of toxicological concern (TTC) in 4 samples.Although the TTC value is intended only for an approximate risk assessment, the obtained results confirm the need for additional toxicological data, exposure and risk assessment regarding the frequent use of herbal food supplements.Furthermore, dietary supplements have become more accessible to consumers, which can lead to extended selfmedication, without consulting a doctor anymore, resulting in significant long-term exposure to any toxin present in dietary herbal supplements.

Figure 1 .
Figure 1.Structural formula of most abundant mycotoxins in HDS.

Figure 4 .
Figure 4.The health benefits stated on the labels of the collected HDS samples.

Figure 6 .
Figure 6.HQ of alternariol monomethyl ether (for samples > LOQ).Red line represents the TTC for an adult of 70 kg bw.

Table 1 .
Validation results of the multi-analyte method.
FOOD ADDITIVES & CONTAMINANTS: PART B

Table 3 .
EDI of mycotoxins and PAs in positive HDS samples and related HQ values.