2.1 Sampling

The analytical data of 2388 food samples from the NFMP from 2015 to 2020 were used in this study. The samples analyzed under the NFMP were collected from different regions in the KSA according to the Standardized Sampling and Transporting Procedures, which are in compliance with the Codex Alimentarius Commission general guidelines on sampling (Codex Alimentarius FAO-WHO), taking into consideration laboratory capabilities. Most (80%) of the food items were collected from the central, western, and eastern regions of the KSA, as these regions represent around 80% of the population distribution in the administrative regions of the KSA. Samples were collected from food factories, storage facilities, and distribution centers (1 kg at least of each sample). The samples were placed in clean and suitable packaging and sent to the SFDA laboratories (Riyadh Laboratory, Jeddah Laboratory, or Dammam Laboratory). Both domestic and imported samples were included in the analysis. Domestic samples are raw commodities that are grown in other countries but are processed, prepared, packed locally by domestic food manufacturers, and sold under Saudi brand names. The imported samples were food commodities that entered the KSA as final products. The sample characteristics are listed in Table 1.

2.2 AFs analytical method

All NFMP samples were analyzed in SFDA laboratories accredited under ISO/IEC 17025 (International Organization for Standardization (ISO). ISO/IEC 17025, 2017). AFB₁, AFB₂, AFG₁, and AFG₂ were quantified using high-performance liquid chromatography (HPLC) with immunoaffinity column cleanup and postcolumn derivatization, as described in ISO 16050:2003 (International Organization for Standardization (ISO)). First, 25 g of each test sample was homogenized with 5 g of sodium chloride (NaCl) (NEN Tech LTD), and 125 mL of water/methanol (H₂O/MeOH) (LiChrosolv) at a ratio of 30/70 (v/v) was added to the homogeneous test sample. The extraction protocol involved the use of an immunoaffinity column specific for AFB₁, AFB₂, AFG₁, and AFG₂. The supernatant was eluted with methanol before injection into the instrument.

The operation conditions of the instrument involve separation using a C18 column (Kinetex 2.6 μ; C18, 100 mm × 3μ) (Phenomenex), with a flow rate of 0.3 mL/min and temperature at 60°C. The instrument was equipped with a fluorescence detector at wavelengths of 362 nm and 455 nm for excitation and emission, respectively, and was equipped with postcolumn derivatization with pyridinium hydrobromide perbromide (PBPB, 0.05 g/1 L) (Sigma-Aldrich) for the determination of AFB₁ and AFG₁.

2.3 Quality assurance

The laboratory used a combination of certified reference materials (CRM), reference materials (RM), quality control samples, and blank samples as tools to ensure the validity of the results. In addition, quality control programs are implemented to ensure the validity of test results, such as participation in proficiency test schemes (PT), calibration programs, preventive maintenance programs, and the use of statistical analysis. The validation parameters included the limit of detection (LOD), limit of quantification (LOQ), linearity, and mean recovery (Table 2).

2.4 Estimation of daily food consumption

Owing to the lack of recent food consumption data in the KSA for use in the assessment of chronic dietary exposure to AFs, the GEMS/Food cluster diets tool developed by The World Health Organization (WHO) was used as the main source to estimate daily food consumption (The World Health Organization). Although these data have many limitations and uncertainties, they are the best available source of food consumption data for estimating dietary exposure to AFs in the KSA. A default value of 0.1 g per day was assigned in the absence of food items in the GEMS/Food cluster diets, as in the case of a food item named halawa tahiniah in the current study.

2.5 Exposure assessment

To achieve a more accurate EDI of AFs and reduce uncertainties, food items with a low number of samples (<5 samples) were grouped with similar food items to better represent the contribution of the food group to the total dietary exposure (e.g., pancake mixes were grouped with desert mixes). Additionally, the outlier results and food items represented by either a low number of samples or for which all data were below the LOD were considered unsuitable for use in the exposure assessment.

To treat left-censored data, the substitution method was applied (European Food Safety Authority, 2010). The lower bound (LB) was obtained by substituting the numerical values of the LOQ with values reported as <LOQ and nondetected samples were assumed to have a value of zero. The upper bound (UB) was obtained by assigning numerical values of the LOQ to all values reported as not detected (ND), <LOD, or <LOQ. The substitution method was only applied to AFB₁, AFB₂, AFG₁, and AFG₂. The sum of all analytical results for the four AFs was calculated to represent the total AFs.

2.6 Risk characterization

Where PHBsAg⁺ and PHBsAg⁻ are carcinogenic potency of AFs for hepatitis B surface antigen-positive (HBsAg⁺) and hepatitis B surface antigen-negative (HBsAg⁻) individuals, respectively.

2.7 Statistical analysis

Statistical analyses were performed using Microsoft Excel 2016. The means and standard deviations (SD) of the LB and UB were determined. A modified Z-score box plot was constructed for the samples to identify outliers.

3.1 Occurrence of AFs

A summary of the concentrations classified by food group is shown in Table 3. The percentage of detected AFs (positive samples) was 12.1% (289 of 2388 samples).

Table S1 illustrates the mean concentrations of all four types of AFs and total AFs in the selected 30 food items. With reference to these food items, the LB mean concentrations of AFB₁ and AFB₂ ranged between ND and 3.8 μg/kg, and ND and 0.62 μg /kg, respectively, whereas the UB mean concentration of AFB₁ and AFB₂ ranged from 0.26 to 3.9 and from 0.05 to 0.64 μg/kg, respectively. In the case of AFG₁ and AFG₂, the LB mean concentrations ranged from ND to 0.34 μg/kg and ND to 0.14 μg/kg, respectively. The UB mean concentration of AFG₁ and AFG₂ ranged from 0.17 to 0.82 μg/kg and from 0.17 to 0.29 μg/kg, respectively.

All four types of AFs were detected in rice, almonds, cashews, hazelnuts, peanuts, peanut butter, and spices, with total AFs ranging from 0.04 to 4.8 μg/kg and from 1.2 to 5.6 μg/kg for LB and UB concentrations, respectively. As expected, AFB₁ was the most frequent AF type found in all food groups except for the spices where the LB and UB for AFG₂ were found to be 1.0 and 1.6 μg/kg, respectively. The LB and UB mean concentrations of AFB₁ in rice were found at 0.15 and 0.36 μg/kg. The highest mean LB and UB contamination of AFB₁ was found in peanut butter samples at 3.8 and 3.9 μg/kg, respectively. The percentage of positive peanut butter samples contaminated with AFs was 82.4% (75 out of 91). Of the positive samples, six were domestic, 55 were imported, and the rest were of unknown origin. This was followed by sunflower seeds and peanuts, for which the percentages of positive samples were 34.8% and 26.2%, respectively. Interestingly, the AFs were ND in cakes, dessert mixes, groats, oats, pasta, popcorn, wheat, wheat flour, walnuts, cowpeas, lentils, and peas. Therefore, these food items were not included in the exposure assessment.

3.2 Exposure assessment

The results of the exposure assessment of AFs in the selected food groups and their contribution to the total dietary intake are presented in Table 4. Owing to the high proportion of left-censored data in the present study, the exposure estimates were likely to be either underestimated or overestimated in the LB and UB scenarios, respectively. The total LB and UB dietary exposures to total AFs for the general population in the KSA through the consumption of the selected food items involved in this study were estimated at 0.29 and 1.84 ng/kg bw/day, respectively. Since AFB₁ is proven to be the most toxic AF type, the EDIs of AFB₁ based on LB and UB scenarios were calculated to be 0.21 and 0.55 ng/kg bw/day, respectively.

The contribution of each food item to the mean EDI of AFB₁ is shown in Table S2. This was calculated using the LB approach to minimize the impact of the LOD and LOQ values. Overall, cereal and grain products contributed the most to the mean chronic dietary exposure to AFB₁ with a mean EDI of 0.178 ng/kg bw/day, accounting for 84.7% of the total LB mean exposure. Rice was the main contributor to the dietary intake of AFB₁ in the cereal and grain groups, with a percentage of 83.3% and mean EDI of 0.175 ng/kg bw/day. This was followed by peanut butter, where the mean EDI of AFB₁ was estimated at 0.01 ng/kg bw/day, accounting for 4.9% of the mean overall EDI of AFB₁.

Moreover, the highest LB and UB mean EDI of AFG₁ were estimated at 0.012 and 0.71 ng/kg bw/day through the consumption of rice. Interestingly, the highest LB mean EDI of AFG₂ was 0.021 ng/kg bw/day in spices, whereas the highest UB mean EDI of AFG₂ was estimated at 0.21 ng/kg bw/day from rice. To sum up, the total LB and UB mean EDI of AFG₁ was 0.016 and 0.89 ng/kg bw/day, respectively. The mean overall EDI of AFG₂ was 0.035 and 0.28 ng/kg bw/day in the LB and UB, respectively.

3.3 Risk characterization

The results of AFB₁ risk characterization are presented in Table 5. The calculated MoE values were 1902.4 and 722.1 for the LB and UB mean EDI of AFB₁, respectively. The estimated cancer risk values following the LB and UB mean EDI of AFB₁ were 0.002 and 0.008 cancer cases/year/100,000 persons, respectively.

4.1 Occurrence of AFs

The AFB₁, AFB₂, AFG₁, and AFG₂ occurrence levels were determined in 2388 food samples. The levels of total AFs in the positive samples were within the limits determined by the Gulf Cooperation Council Standardization Organization (GCC Standardization Organization) except for two rice samples and five pistachio samples, three of which were imported, and the others were of unknown origin. The results showed a difference in AFB₁ mean concentrations between peanuts and their product (peanut butter) where the levels found at 0.31 and 3.7 μg/kg, respectively, indicated that processing methods and storage conditions are very critical points to be monitored and controlled by food producing companies. Remarkably, AFB₁ mean concentrations in peanut butter samples collected from Riyadh, Jeddah, and Dammam were detected at the same level of 3.7 μg/kg. This could be because peanut butter is considered a processed food that is distributed throughout the KSA under the same conditions, unlike rice where the AFB₁ mean concentrations in samples collected from Jeddah and Dammam were considerably higher than those collected from Riyadh (0.27 and 0.07 μg/kg, respectively). This might be because humid climates in the eastern and western regions provide an optimal environment for fungal growth in raw materials, such as rice.

This study showed that nuts and seeds are the major food groups contaminated with AFs in the KSA. Our findings are in agreement with those of a recent study conducted in the KSA by Serdar et al. (Serdar et al., 2020), which measured AF levels in different food groups. However, their results were considerably higher than those obtained in our study, with mean total AF concentration values of 282.28 μg/kg in walnut samples and 12.7 μg/kg in pistachio samples. This could be due to the fact that all samples in their study were taken from the western region, which is known for its high temperature and humid weather most seasons of the year. Furthermore, El Tawila et al. (El Tawila et al., 2020) found that 27.3% and 20% of nut samples (n = 99) were contaminated with AFB₁, AFB₂, AFG₁, and AFG₂, respectively, which was much higher than the percentages of positive samples in our study (14%). The variations in concentrations between the two studies could be due to the nut moisture content, soil properties, and poor storage conditions, resulting in fungal growth. Ashraf (Ashraf, 2012) studied the occurrence of AFs in dry nuts consumed in the KSA and found that AFB₁ was detected in 92% of cashew nuts, six of them exceeding the ML of 5 μg/kg set by the EU, whereas a lower contamination level was found in macadamia samples (~85% of the samples). In contrast, Bokhari (Bokhari, 2001) determined the levels of AFs in various food groups, such as cereal grains, spices, nuts, beans, and oilseeds, in the KSA. Similarly, the highest AFB₁ levels were detected in oilseeds and cereal grains with mean concentrations of 12.9 and 8.02 μg/kg, respectively; however, they found lower levels in the beans and nuts (mean concentrations of 1.99 and 2.73 μg/kg, respectively). Surprisingly, another recent study measured the levels of the four AFs and the total AFs in rice, with all samples below the LOQ values (Almutairi et al., 2021). They justified the absence of AFs in nearly 480 samples with the overuse of fungicides by farmers, which could have affected the growth of the fungus in rice samples (Kumar et al., 2013).

On an international level, studies have measured the occurrence of AFs in food commodities (Cano-Sancho et al., 2013; Ding et al., 2012; Hassan et al., 2022; Kang'ethe et al., 2017; Sugita-Konishi et al., 2010; Thuvander et al., 2001; Zinedine et al., 2007). The total AF mean concentrations observed in the present study were much lower than the total AF concentrations reported in Spain, with a mean level of 2.7, 8.9, 0.5, and 0.9 μg/kg in peanuts, pistachios, breakfast cereals, and corn, respectively (Cano-Sancho et al., 2013). In Sweden, the levels of total AF contamination were extremely high in pistachios and Brazilian nuts, with maximum values of 2200 and 2500 μg/kg, respectively (Thuvander et al., 2001). Additionally, very high levels of AFs have been reported in Morocco, where 50% of corn samples were contaminated with AFB₁ at a mean concentration of 1930 μg/kg (Zinedine et al., 2007). In the same study, all red paprika samples (100%) were contaminated with AFB₁ at a mean concentration of 2.88 μg/kg. In Lebanon, AFB₁ was detected in all 105 collected white, parboiled, and brown rice samples with a mean AFB₁ concentration of 0.5 μg/kg and a range of 0.06–2.08 μg/kg (Hassan et al., 2022), while the present study found lower levels of AFB₁ in rice, with mean concentrations at 0.15 and 0.36 μg/kg in the LB and UB, respectively, as stated earlier in the study's results.

4.2 Exposure assessment

Although the highest mean AFs concentrations were found in the nut and seed groups, cereals and grain products made the largest contribution to the mean EDI of AFs. Moreover, cereal and grain products accounted for more than 80% of the overall mean EDI of AFB₁. This might be due to the high consumption of rice (82.3 g per day) as one of the main staple foods in the KSA. In Japan, the EDI of AFB₁ ranges from 0.003 to 0.004 ng/kg bw/day, whereas the EDI of total AFs ranges from 0.003 to 0.014 ng/kg bw/day, considering chocolate and peanut butter as the main contributors to the EDI of total AFs. This study concludes that the current dietary intake of AF in Japan is low and has no considerable effect on human health (Sugita-Konishi et al., 2010). Another study assessed the EDI of the Catalonian population for AFs through the consumption of peanuts, pistachios, dried figs, sweet corn, breakfast cereals, and corn snacks (Cano-Sancho et al., 2013). Their results illustrated that peanuts and pistachios were the highest contributors to the total AF intake in adult males and females, with a mean EDI of total AFs of 0.135 and 0.078 ng/kg bw/day, respectively. Additionally, another study examined 300 samples of hazelnuts and dried figs to estimate the EDI of AFB₁ and total AFs in the Turkish population. The study found that the EDI of AFB₁ and total AFs ranged from 0.003 to 0.016 ng/kg bw/day and from 0.004 to 0.023 ng/kg bw/day, respectively, and the contribution of hazelnuts to the EDI of total AFs was higher than that of dried figs (Kabak, 2016). By comparing the estimated total AFs intake of the Saudi adult population in the current study with the ranges mentioned above, it can be concluded that the EDI of total AFs in the KSA is higher, regardless of the lower concentration values. Nevertheless, Andrade et al. (Andrade et al., 2013) estimated considerably higher EDI values for the Brazilian population, with estimates ranging from 6.8 to 27.6 ng/kg bw/day, indicating a potential health concern. According to a report conducted by Shephard (Shephard, 2008) that examined AFs in Africa, the intake of the Kenyan population from consuming 400 g per day of maize was estimated to be 133 ng/kg bw/day, whereas the intake from peanuts was reported at 99.2 ng/kg bw/day in Ghana (Williams et al., 2004). In a previous study conducted in China, in which 1854 samples were analyzed to assess the probabilistic risk of dietary intake of AFB₁ contamination in different food groups, the results revealed that rice and rice products were the main contributors to the dietary intake of AFB₁. The EDI of AFB₁ in the adult group was estimated at 0.59 ng/kg bw/day (Zhang et al., 2020).

In Serbia, the mean overall EDI of AFB₁ among male and female adults was in the range of 0.56–0.76 ng/kg bw/day and 0.52–0.72 ng/kg bw/day for the LB and UB scenarios, respectively (Udovicki et al., 2021). These values are higher than the results reported in the present study. Table S2 illustrates the results of the exposure assessment for the four types of AFs and total AFs, taking into account the available consumption data.

4.3 Risk characterization

AFB₁ is the most toxic AF type and is most frequently found in food commodities, whereas the other three types are rarely found in the absence of AFB₁ (EFSA Panel on Contaminants in the Food Chain (CONTAM) et al., 2020). To conduct risk characterization estimates, the MoE approach was adopted to evaluate the risks related to human health associated with AFB₁. As identified by the EFSA in their report on the risk assessment of genotoxic and carcinogenic substances, a small MoE indicates a higher risk than a larger MoE, where values below 10,000 represent a human health concern (European Food Safety Authority (EFSA), 2005). In this study, the estimated MoE values were 1902.4 and 722.1 for the mean LB and UB EDI of AFB₁, respectively. By comparing the MoE range in the present study with the range provided by Shephard (Shephard, 2008), it can be anticipated that foods consumed in the KSA are less contaminated than those consumed in Africa, where most MoE values are below 10. In addition, our MoE value is considerably higher than that reported in Brazil, where the MoE value was estimated at 25 for the total population, indicating a potential risk for consumers, although the variety of food items collected and analyzed in this study was wider than that reported by Andrade et al (Andrade et al., 2013). Moreover, the MoE values estimated in a study conducted in China ranged from 24.1 to 1273 based on the AFB₁ exposure estimate from the consumption of peanuts (Ding et al., 2012). As reported by Kabak (Kabak, 2021), the calculated MoE values ranged from 995 to 860 for adult Turkish consumers, based on the mean EDI of AFB₁, which was assessed in five food groups. The above-mentioned MoE ranges, including the calculated values in the current study, were all far lower than 10,000, which could be indicators of health concerns in long-term exposure. It should be emphasized that the number of food items investigated in this study was greater than that in previous studies, which resulted in higher exposure estimates and therefore, lower MoE values.

Epidemiological studies suggest a correlation between AFB₁ exposure and an increased incidence of primary liver cancer in the presence of other risk factors such as HBV infection (Liu et al., 2012). The estimated cancer risk values in the present study varied greatly when compared to the estimations from the EFSA's report on risk assessment of AFs in foods, where the cancer risk ranged between 0.004 and 0.057 cancer cases/year/100,000 persons for adults following LB and UB mean EDI of AFB₁ (EFSA Panel on Contaminants in the Food Chain (CONTAM) et al., 2020). Moreover, the potential cancer risk at the 90th percentile in Japan is even lower, with estimates ranging from 0.00004 to 0.00005 cancer cases/year/100,000 persons (Sugita-Konishi et al., 2010). In contrast, a very high population risk was estimated for the Ghanaian population, where it reached 70.1 cancer cases/year/100,000 persons (Shephard, 2008). Similarly, based on the average dietary intake of different food commodities contaminated with AFs in Bangladesh, the associated risk of HCC was estimated to be 1311 cancer cases/year/100,000 people, indicating a high risk to the population (Saha Turna & Wu, 2019). In conclusion, the estimated cancer risks in the general population following exposure to AFB₁ were consistent with the conclusions drawn from the MoEs.

There are some shortcomings in the present study that should be addressed in future research. First, the exposure assessment was performed for the adult population as a single group, without considering sex-based differences in food consumption patterns, weight status, and dietary habits. It is also important to note that a national food consumption database should be considered to provide a more realistic risk assessment. However, owing to the lack of these data in the KSA, an international food database was used, which may have either underestimated or overestimated the EDI of AFs and, consequently, the risk characterization.