Risk of dietary and breastmilk exposure to mycotoxins among lactating women and infants 2–4 months in northern India

Abstract Mycotoxins are carcinogenic secondary metabolites of fungi that have been linked to infant growth faltering. In this study, we quantified co‐occurring mycotoxins in breast milk and food samples from Haryana, India, and characterized determinants of exposure. Deterministic risk assessment was conducted for mothers and infants. We examined levels of eight mycotoxins (Aflatoxin B1, B2, G1, G2, M1, M2; Ochratoxin A, B) in 100 breast milk samples (infants 2–4 months) using ultra‐high‐performance liquid chromatography tandem mass spectrometry. Aflatoxin B1 (AFB1), fumonisin B1 (FB1) and deoxynivalenol (DON) were detected in several food items (n = 298) using enzyme‐linked immunosorbent assays. We report novel data on the presence of mycotoxins in breast milk samples from India. Whereas breast milk concentrations (AFM1 median: 13.7; range: 3.9–1200 ng/L) remain low, AFM1 was detected above regulatory limits in 27% of animal milk samples. Additionally, 41% of infants were above provisional maximum tolerable daily intake (PMTDI) limits for AFM1 due to consumption of breast milk (mean: 3.04, range: 0.26–80.7 ng kg−1 bw day−1). Maternal consumption of breads (p < 0.05) was associated with breast milk AFM1 exposure. AFB1 (μg/kg) was detected in dried red chilies (15.7; 0–302.3), flour (3.13; 0–214.9), groundnuts (0; 0–249.1), maize (56.0; 0–836.7), pearl millet (1.85; 0–160.2), rice (0; 0–195.6), wheat (1.9; 0–196.0) and sorghum (0; 0–63.5). FB1 (mg/kg) was detected in maize (0; 0–61.4), pearl millet (0; 0–35.4) and sorghum (0.95; 0–33.2). DON was not detected in food samples. Mothers in our study exceeded PMTDI recommendations for AFB1 due to consumption of rice and flour (mean: 75.81; range: 35.2–318.2 ng kg−1 bw day−1). Our findings show the presence of Aflatoxin B1 and M1 at various levels of the food chain and in breast milk, with estimated intakes exceeding PMTDI recommendations. Aflatoxins are known carcinogens and have also been linked to stunting in children. Their presence across the food system and in breast milk is concerning, thus warranting further research to replicate and expand on our findings and to understand implications for maternal and child health.


| INTRODUCTION
Mycotoxins are secondary metabolites of fungi and are found in 25% of the global food system, resulting in annual losses of over 1 billion metric tons of food and feed products (Smith et al., 2016;Wu et al., 2014). In addition to being known carcinogens and genotoxins, literature suggests that mycotoxins may also be associated with infant growth faltering and stunting (Tesfamariam et al., 2019). Mycotoxins are found in a variety of food commodities such as maize, wheat, rice and animal source products, in addition to breast milk (Bayman & Baker, 2006;Deepa & Sreenivasa, 2017;Shirima et al., 2013;Smith et al., 2016;Sobrova et al., 2010;Wild & Gong, 2010). Thus, breast milk is a potential source of dietary exposure for infants and young children (Coppa et al., 2019;Fakhri et al., 2019). There are hundreds of different mycotoxins, but aflatoxins (AFs), ochratoxins (OTs), deoxynivalenol (DON) and fumonisins (FUM) are of particular consequence to public health (Warth et al., 2016).
Aflatoxins are produced by Aspergillus flavus and Aspergillus parasiticus fungi, and AFB 1 is the most prevalent and toxic metabolite, classified as a group 1 carcinogen, known to cause cancer in humans (IARC [International  Agency  for  Research  on Cancer], 1993, 2012). Ochratoxins are produced by Aspergillus ochraceus and Aspergillus penicillum (Bayman & Baker, 2006;Smith et al., 2016). OTA is classified as a group 2B agent, possibly carcinogenic to humans (IARC, 1993). Deoxynivalenol is produced by Fusarium graminearum and Fusarium culmorum and is classified as a group 3 agent, with inadequate evidence for human carcinogenicity (IARC, 1993). Fumonisins are another important group of mycotoxins produced by Fusarium verticillioides (Fusarium moniliforme) and found predominantly as FB 1 (70%) in commodities (Deepa & Sreenivasa, 2017). Fumonisins are classified as group 2B carcinogens (IARC, 1993(IARC, , 2002. Breast milk contains elements essential for child health, growth and development (World Health Organization, 2019). During early infancy, the primary exposure source to mycotoxins is breast milk, where lactational transfer of mycotoxins occurs via maternal diet (Warth et al., 2016). AFM 1 , also known as milk aflatoxin, is a hepatic hydroxylation by-product of AFB 1 . It is tenfold less toxic and, although previously classified as a group 2B carcinogen, has in more recent IARC monographs been re-evaluated to a group 1 carcinogen (IARC, 1993(IARC, , 2002(IARC, , 2012. AFM 1 and OTA have been detected in breast milk samples from across the globe (Coppa et al., 2019;Fakhri et al., 2019). Infants and young children may experience the adverse effects of mycotoxin exposure up to three times greater than adults because of their larger intake/body weight ratio, higher metabolic rate and lower capacity to detoxify these contaminants (Assuncao et al., 2015;Hulin et al., 2014;Sherif et al., 2009).

Few studies have linked mycotoxin concentrations in breast milk
with maternal diet to understand the role of the food system as a source of exposure (Galvano et al., 2008;Ortiz et al., 2018). To the best of our knowledge, no studies from India and only one from the South Asia region have reported on mycotoxin concentrations in breast milk (Khan et al., 2018).
Therefore, the aim of this cross-sectional observational study was to quantify co-occurring mycotoxins in an assortment of commonly consumed cereal crops (AFB 1 , FB 1 and DON), commercial infant formula (AFB 1 ) and 100 breast milk samples (AFs B 1 , B 2 , G 1 , G 2 , M 1 , M 2 , and OTs A, B) from rural and peri-urban communities in Haryana, India. These data were used to understand determinants of breast milk mycotoxin exposure, characterize risk among mothers and children and explore variations in levels of mycotoxins across seasons (over the period of a year) and locations (peri-urban vs. rural).

| Sample preparation
Aliquoted breast milk samples (1 ml) were allowed to thaw on top of ice and transferred to 15 ml centrifuge tubes. Ice-cold ACN (2 ml) with 2% formic acid was then added, and 10 μl of internal standards were spiked on top. The sample was vortexed for 1 min and kept on ice for 5 min to allow for complete protein precipitation. Next, 300 μl of concentrated ammonium acetate (10 g/ml) solution was added. The sample was sonicated in a water bath sonicator for 10 min and kept on ice for 5 min and then centrifuged (5 min, 3800×g) to separate the ACN layer.
Approximately 1.5 ml of ACN from the supernatant was transferred to 2-ml Eppendorf tube and dried in a speed vacuum. The final reconstitution was done with 100 μl of 50% ACN and filtered through a 0.45-μm PVDF membrane centrifugal filter (5 min, 1500×g). The top 80 μl from each sample was transferred to HPLC vials and 10 μl from each was injected for analysis of mycotoxins in milk using the UHPLC-MS/SRM method. A seven-point calibration curve (rangesof 15.6-1000 ng/L for AFB 1 , AFB 2 , AFG 1 and OTA; 7.8-500 ng/L for AFM 1 , AFM 2 and OTB and 78-5000 ng/L for AFG 2 ) was prepared on a daily basis, in addition to quality controls at limit of quantification (LOQ), low, medium and high-quality control levels. All animal milk samples were analysed using UHPLC-MS/MS in a manner similar to the breast milk samples, described above.

| Instrumental analysis
Sample analysis was conducted using a Sciex QTRAP 6500 (Sciex Singapore) mass spectrometer, with a turbo V ion source. The mass spectrometer was coupled to an Agilent 1290 infinity II UHPLC system (Agilent Technologies India Pvt. Ltd., India) and equipped with a column oven (set to 40 C), an auto-sampler and thermo-controller (set to 10 C). Mobile phase solvent A was water (10-mM ammonium acetate, 0.1% formic acid) and solvent B was ACN (0.1% formic acid).
A C-18 column (2.1 × 100 mm, 1.8 μm, Agilent, Inc.) was used for separation of mycotoxins. We used an optimized gradient to achieve We further optimized DP, CXP and CE for each intense product ion (Table S2).
2.5 | Food sample AFB 1 , FB 1 and DON analysis 2.5.1 | Sample preparation Food samples were analysed using direct and indirect competitive enzyme-linked immunosorbent assays (ELISA). Approximately 100 g of each food sample was ground to a fine powder using a Kenstar Senator blender (Kenstar, Gurgaon, India). Next, 100 ml of 70% methanol (v/v-70 ml absolute methanol in 30-ml distilled water) containing 0.5% KCl was added to 20 g sample powder in an Erlenmeyer flask.
For DON extraction, 100-ml deionized H 2 O was used in place of methanol, in accordance with the kit manufacturer's protocol. Extracts were incubated at room temperature for 60 min on a revolving shaker (250 rpm), filtered through Whatman no. 4 filter paper into a fresh tube and stored at 4 C until ELISA analysis. A similar protocol was used to prepare a toxin-free sample extract (healthy groundnut), which served as a realistic crop matrix used for dilution of AFB 1 and FB 1 standards and as a negative control.

| ELISA assay procedures
For AFB 1 and FB 1 , we conducted indirect competitive ELISA according to the protocols developed by ICRISAT (Reddy et al., 2001).
Between each step of the protocol, contents were decanted and the plate washed three times with a PBST wash buffer. ELISA plates were coated with 150-μl AFB 1 -bovine serum albumin (BSA; 100 ng/ml) for AFB 1 ELISAs, or FB 1 -BSA (500 ng/ml) for FB 1 ELISAs, both prepared in carbonate buffer (100 ng/ml) and incubated at 37 C for 1 h.
Following this, blocking was conducted by adding PBST to each well and incubating at 37 C for 30 min. Standards (AFB 1 : 25-0.097 ng/ml; FB 1 : 6-0.047 μg/ml) were prepared in 10% toxin-free extract with 7% methanol, and 100 μl was added to each well of the plate. Next, 100 μl of diluted sample extract (1:10 in PBST-BSA) and 50 μl of anti-

| Dietary exposure assessment and risk characterization
Estimated daily intake (EDI) of AFB 1 and AFM 1 for mothers via consumption of a variety of food items and animal milk and of AFM 1 for infants through breast milk were calculated using a deterministic exposure assessment approach (Assuncao et al., 2015;Cantu-Cornelio et al., 2016;Ortiz et al., 2018). EDIs were calculated using the equation below and reported as ng kg −1 bw day −1 for relevant mycotoxins in each food item for which consumption data were available (Ortiz et al., 2018).  -Goodman, 1995-Goodman, , 1998. The extent to which exposure exceeds these values was used to assess risk (Cunha et al., 2018).
The European Food Safety Agency (EFSA) has also proposed the use of margin of exposure (MOE) for risk analysis of aflatoxins. MOE represents the ratio between BMDL 10 (benchmark dose lower confidence limit for 10% extra risk of liver tumour formation in rats amounting to 170 ng kg −1 bw day −1 ) and EDI (EFSA, 2005). Here, an EDI is considered of concern to public health if the MOE is lower than 10,000. MOE values do not quantify risk but are used to indicate a level of concern. A lower MOE indicates a higher level of concern (EFSA, 2005).

| Population characteristics
Sociodemographic characteristics of households, in addition to details about mothers and children in our sample, are presented in Table 1.
Contamination levels for AFM 1 in milk ranged between 0 and 1200 ng/L. Only one sample was detected with AFM 1 levels above Food Safety and Standards Authority of India (FSSAI) set limits in animal milk (500 ng/L) ( Table 2). There are currently no cut-offs for aflatoxins in breast milk samples as these are not commercially regulated.

Details of individual food items composing each group in our
FFQs and median frequencies of daily and weekly intake are presented in Table S4. TOBIT regression analyses of the determinants of AFM 1 exposure in breast milk samples showed maternal consumption of items such as tandoori roti and stuffed parathas, both wheat flour based Indian flatbreads to be significantly (p < 0.05) associated with an increase in breast milk concentrations of AFM 1 . This trend remained significant for both bread items after adjusting for maternal age and household wealth index ( (Continues) procurement of staple crops, self-reported instances of insect infestation and moisture in household storage of crops).
Concentrations of all mycotoxins examined in breast milk by season are shown in Figure 1. We did not see statistically significant seasonal differences in values of AFM 1 in breast milk samples across the 12-month span for which data were collected (Figure 2). Mean concentration of OTA in our study was 23.15 (±7.78) ng/L, and only two samples were above the limit of quantification. Other mycotoxins were detected at levels less than the LOQ in our study and have therefore not been considered for further statistical analyses. Median (IQR) concentrations of mycotoxins in breast milk are presented in Table 2.  including AFB 1 (n = 19), AFB 2 (n = 23), AFG 1 (n = 1) and AFM 2 (n = 23) were seen in several samples at levels above the limit of detection ( Figure S2). OTA and OTB were not detected in any of the animal milk samples in this study.

| Occurrence of Aflatoxin B 1 , Fumonisin B 1 and DON in food samples
Aflatoxin B 1 was detected across a range of food items at all three seasonal time points, and several food samples were above FSSAI set regulatory limits. Median sample AFB 1 concentrations by food items are presented in Table S3. We saw statistically significant (p < 0.05) variations in AFB 1 levels across seasons for wheat, sorghum, flour, groundnuts, rice and pearl millet, with highest mean concentrations observed during the monsoon collection period (Tables S1a and S1b).
Negligible levels of AFB 1 were detected in 10 commercial infant formula samples collected in our study.
Fumonisin B 1 was analysed only in maize, pearl millet and sorghum, given the known vulnerability of these commodities to infestation by implicated fungi. The toxin was detected in these commodities across the three seasonal time points (Table S1). Samples were above regulatory limits set by the European Union for maize (0.2 mg/kg) at all three time points. We saw statistically significant seasonal trends for mean concentrations of FB 1 in pearl millet and sorghum samples. Both pearl millet and sorghum are harvested in the fall, which likely explains non-detected levels of FB 1 contamination during the fall collection period. We did not see any statistically significant trends for variation in mean concentrations of AFB 1 or FB 1 among rural and peri-urban sites and between types of markets, namely, village, mid-retail and wholesale levels. Deoxynivalenol was not detected in any of the food samples in our study, potentially owing to low disease pressure by F. graminearum and the generally warm climate in the study region.

| Deterministic exposure assessment and risk assessment
Maternal exposure assessment to mycotoxins through intake of animal milk, flour and rice was conducted. Data from the 24-h recall show that 93% of women in our sample consumed milk at an average volume of 0.33 L/day (±0.30), 92% consumed wheat flour at 0.24 kg/day (±0.10) and rice was consumed by 21% of our sample at 0.19 kg/day (±0.14). Mean maternal estimated daily intake of AFM 1 due to milk was 3.58 (range: 0.42-17.74) ng kg −1 bw day −1 . Mean estimated daily intake for AFB 1 via flour was 82.3 ng kg −1 bw day −1 and 46.4 ng kg −1 bw day −1 via rice (Table 4).
Using a deterministic risk assessment approach, 80% of women exceeded this cut-off for AFM 1 due to milk consumption; 100% of our sample significantly exceeded the PMTDI level of 1 ng kg −1 bw day −1 for AFB 1 due to consumption of rice and wheat flour (Kuiper-Goodman, 1998;Yogendrarajah et al., 2014). MOE values for maternal aflatoxin exposure from flour (range: 0.90-17.21), rice (range: 1.33-17.54) and milk (range: 9.58-404.53) were also lower than 10,000 indicating that exposure to AFB 1 and AFM 1 in the food system are of priority for risk management actions (Ortiz et al., 2018).
Mean estimated daily intake value of AFM 1 from breast milk for infants in our study was 3.04 ng kg −1 bw day −1 , 41% of the sample exceeded PMTDI levels with upper percentiles of milk consumption exceeding thresholds by 3-80 folds (P90-P99). Approximately 16% of infants in our sample consumed animal milk, 7% consumed formula and 5% received food grain and/or porridge made of grain, in addition to breast milk. Thus, exposure to mycotoxins in dairy milk and via contaminated complementary foods may be of concern among children in this population.

| DISCUSSION
We Samples from our study show concentrations to be within the ranges described by others, although no other studies on breast milk mycotoxins from India currently exist in the literature. The overall prevalence of AFM 1 in breast milk from studies conducted globally is highly variable and associated with mycotoxin contamination in food items and in turn maternal dietary patterns, which differ across countries (Abdulrazzaq et al., 2003;Cherkani-Hassani et al., 2016;Coulter et al., 1984;Memis & Yalcin, 2019).
Our findings suggest that the consumption of breads and rotis among women in this population is associated with increased concentrations of AFM 1 in milk. This is likely due to the presence of AFB 1 in flour used to produce these Indian flatbreads and rotis. As an important caveat, we acknowledge that associations noted here may be spurious and larger sample sizes are needed to validate our findings.
Carry-over of AFB 1 from diet and excretion as AFM 1 in breast milk is estimated to be between 0.1% and 0.4% (Zarba et al., 1992).
Prior studies have found maternal consumption of cereals, peanut butter, vegetable oil, rice (Elzupir et al., 2012), cow milk (Mahdavi et al., 2010) and sausage (Jafarian-Dehkondi & Pourradi, 2013) to be associated with increased concentrations of AFM 1 in breast milk from women in Iran. A study conducted in Italy found that lactating women with high AFM 1 (140 ng/L) in breast milk had consumed a large amount of corn meal-based foods in substitution for cereal-based food such as rice, pasta, bakery products and breakfast cereals (Galvano et al., 2008). Consumption of corn oil, peanuts, raw milk, beans and wheat meal have been associated with higher breast milk AFM 1 in Egypt and Nigeria (Adejumo et al., 2013;El-Tras et al., 2011;Polychronaki et al., 2006). Higher consumption of rice and chocolate have also been associated with AFM 1 in milk (Bogalho et al., 2018).
We did not see any  (Adejumo et al., 2013;Bogalho et al., 2018;Polychronaki et al., 2006). Bogalho et al. (2018) noted that breast milk collected at early stages of lactation featured mean concentrations of AFM 1 higher than samples collected 6-12 months post partum. Several additional determinants of lactational transfer of mycotoxins have been identified, including but not limited to maternal dietary diversity and hydration, frequency of infant feeding and breast infection (Warth et al., 2016).
There remains a dearth of knowledge about the lactational transfer rates of mycotoxins from blood to breast milk. Milk to plasma ratios for AFM 1 were estimated to be 0.21 among nursing Egyptian mothers (Hassan et al., 2006), although this metabolite is transient in plasma, and concentrations of AFM 1 in breast milk are higher at earlier stages of lactation (Degen et al., 2013;Polychronaki et al., 2006). We Our findings for AFB 1 in food items are in alignment with others that have documented levels of aflatoxins in maize, groundnuts, spices, peanuts, corn, rice, soybeans, sorghum, cereals, and chilies at levels ranging between 0 and 46,000 μg/kg (Reddy et al., 2010).
Higher levels of AFB 1 in food samples collected during the monsoon season in our study are likely due to increased moisture levels and humidity, known to impact the growth of mycotoxigenic mould (Ojuri et al., 2019). A large multicentre study conducted in India found 40.3% of wheat samples collected from across the country to have AFB 1 levels ≥5 μg/kg, with 16% of samples over the Indian permissible regulatory limit of 30 μg/kg, when that study was published (current legal limit in India is 15 μg/kg) (Toteja et al., 2006). Additional studies of items intended for animal feed from India have found aflatoxin contamination in groundnut cake, maize, millets, rice bran, sorghum, soybeans, sunflower and mixed feeds in excess of 10 μg/kg (Thirumala-Devi et al., 2002). in these items, in addition to poultry feed from India (Shetty & Bhat, 1997 Although limited by a small sample size, we have used gold standard methods for the detection and quantification of mycotoxins in breast milk samples. We were also able to assess seasonality in relation to concentrations of mycotoxins at various levels of the food ecosystem and from breast milk within study communities.
We were not able to calculate estimated daily intake values for all foods consumed by mothers in our sample. It is therefore likely that cumulative exposure is higher than noted here, from other sources, in addition to exposure to other mycotoxins not quantified in our study.
It is also critical to investigate mycotoxin exposure among infants who are not being exclusively breastfed, and those consuming weaning foods, as exposures are likely to be higher in these groups. Studies in East Africa have shown exposure to AFs and FBs in infant foods to be of concern among non-exclusively breastfed children under 6 months of age (Magoha et al., 2016). Another limitation of this study is that the matrix effects of different commodities in the validated AFB 1 and FB 1 ELISA protocols have not been thoroughly explored. While it is unlikely that accounting for commodity matrix effects in food samples would qualitatively change our results or conclusions, evaluating and accounting for such effects may yield more precise estimates in future work.
Finally, there is a need for longitudinal cohort studies, complemented with total diets data to examine exposure to mycotoxins across the different stages of lactation, particularly in light of mixed evidence regarding their effects on birth outcomes including smallfor-gestational age (SGA), infant growth faltering and stunting

| CONCLUSION
Breast milk remains the best source of essential nutrients and immune factors for infants, and exclusive breastfeeding is the recommended standard of practice for children under 6 months (World Health Organization, 2019). Although AFM 1 was detected in 41% of breast milk samples, concentrations of the aflatoxin remain low and below FSSAI set regulatory limits. Exposure assessments suggest levels of aflatoxins in food are higher than permissible limits. Women who consume flour-based products such as bread are at increased risk of mycotoxin exposure, in this population.
Between 80% and 100% of women in our sample had dietary intakes of AFs above recommended daily limits of exposure. Given a growing body of evidence showing dose-dependent associations between the consumption of aflatoxins and liver cancer, other mycotoxins and cancers of the breast and cervix, more in-depth evaluations of mycotoxin exposure and contamination across the food system, from feed to human, are prudent and timely (Claeys et al., 2020). Furthermore, human epidemiological studies remain the need of the hour to support evidence-based public health strategies to mitigate mycotoxin contamination and their harmful effects on all humans, with an emphasis on vulnerable populations including mothers and children.

ACKNOWLEDGMENTS
We would like to acknowledge the field teams at CHRD-SAS for their relentless efforts. We would like to thank Sweekruthi Shetty at

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.