Multiple Mycotoxins in Rice: Occurrence and Health Risk Assessment in Children and Adults of Punjab, Pakistan

Mycotoxin contamination in rice can create a health risk for the consumers. In this study, the measurement of 23 mycotoxins in rice samples (n = 180) was performed using a validated LC–MS/MS method. A food frequency questionnaire was used to get rice consumption data for the assessment of mycotoxin dietary exposure, before calculating the health risk in adults and children of north and south regions of the Pakistani Punjab province. The prevalence of aflatoxin B1 (56%), aflatoxin B2 (48%), nivalenol (28%), diacetoxyscirpenol (23%), fumonisin B1 (42%), zearalenone (15%), HT-2 toxin (10%), deoxynivalenol (8%), and ochratoxin A (6%) was estimated in samples with a mean concentration range between 0.61 and 22.98 µg/kg. Aflatoxin degradation by traditional Pakistani cooking recipes was evaluated and observed to be 41–63%. The dietary exposure to aflatoxins exceeded the tolerable daily intake at all levels, and ochratoxin A and zearalenone posed health risk at high contamination and high consumption levels. The margin of aflatoxin B1 exposure ranged between 10 and 69 in adults and 10 and 62 in children. The mean cancer risk by aflatoxin B1 exposure was 0.070 (adults) and 0.071 (children) cases/year/100,000 people in South Punjab population, and 0.122 (adults) and 0.127 (children) cases/year/100,000 people in North Punjab population. This study will provide new insights for the planning and management of mycotoxins in Pakistan.


Introduction
Mycotoxins are naturally occurring toxic secondary metabolites produced by fungal species of the genera Aspergillus, Alternaria, Penicillium, Fusarium, Claviceps, and several others. Mycotoxins can be nephrotoxic, immunosuppressive, carcinogenic, and teratogenic. Trichothecenes, aflatoxins (AFs), Alternaria toxins, fumonisins (FBs), ochratoxin A (OTA), and zearalenone (ZEN) are the most important aflatoxin/kg b.w. day) [22]. The second approach calculates the margin of exposure (MoE), which is the ratio between the toxicological threshold derived from animal studies and the estimated exposure in humans [23].
Limited data is available on the determination of and dietary exposure to multi-mycotoxins in Pakistani rice. Previous studies embarked upon the occurrence and dietary exposure to only a few mycotoxins (AFs, OTA, and ZEN) [25,26,30,31]. These studies did not include the dietary exposures to multiple mycotoxins, especially in children, are devoid of important parameters such as the calculation of the margin of exposure (MoE), and did not include a food frequency questionnaire (FFQ) and the degradation effect of Pakistani cooking recipes on the prevalent mycotoxins.
To the best of our knowledge, no study has been published on the simultaneous determination of multi-mycotoxins covering aflatoxins (AFB 1 , AFB 2 , AFG 1 , AFG 2 ), OTA, fumonisins (FB 1 and FB 2 ), DON, nivalenol (NIV), diacetoxyscirpenol (DAS), HT2, and ZEN in Pakistani rice samples as well as on their dietary exposure in adults and children of Pakistan using confirmatory analytical techniques. Hence, to fill the aforementioned gaps, this work aimed to study the contamination levels of multi-mycotoxins in rice and their exposure risk assessment, based on rice consumption data in different regions. After accounting for the degradation effect of Pakistani cooking recipes on the prevalent mycotoxins, the study subsequently enabled us to determine the dietary exposure to mycotoxins in two population groups (adults and children), as the intake exceeding the provisional maximum tolerable daily intakes (PMTDIs) set by the European Food Safety Authority (EFSA) that poses health risks. Also, the potential risk of liver cancer arising from aflatoxin B 1 intake was assessed in the Pakistani population of South Punjab (SP) and North Punjab (NP) regions. The study is likely to contribute to devising strategies for the planning and management of these food contaminants for the Pakistani Punjab population.

Validation
The validation parameters were calculated from the relative peak area, using deepoxy-deoxynivalenol (DOM) as an internal standard for NIV, DON, fusarenon-X (FUS-X), neosolaniol (NEO), 3-acetyl-deoxynivalenol (3-ADON), 15-acetyl-deoxynivalenol (15-ADON), HT2, T2, and DAS, while zearalanone (ZAN) was used for the rest of the analytes. The validation parameters were calculated in terms of linearity, limit of detection (LOD), limit of quantification (LOQ) ( Table 1), % apparent recovery, intraday repeatability, interday repeatability, and expanded measurement uncertainty (Table A1). Table 1. Limits of detection (LOD) and limits of quantification (LOQ), linearity range, and coefficient of determination (R 2 ) of all the measured mycotoxins in rice. There were no interfering peaks at the retention times of all the analytes. The specificity was checked by running 20 blank rice samples on LC-MS/MS. The blank rice samples were spiked at five calibration levels with each compound, and the calibration curves were prepared each day in triplicate for four validation days. For the samples in which the contamination level did not fall in this validated range, either lower or higher range calibration curves were plotted to quantify them. The method-matched calibration curves fitted by linear regression showed a coefficient of determination (R 2 ) ranging from 0.991 to 0.999. The mean recoveries for all the analytes were in the range of 80% to 111% (Table A1). The relative standard deviation was calculated in intraday repeatability (RSD r ) and interday repeatability (RSD R ) conditions and ranged from 3% to 14% and from 6% to 20%, respectively. The LOD values of all the mycotoxins ranged from 0.5 to 53 µg/kg. The LOQs for all the toxins ranged from 1.5 to 105 µg/kg. The expanded measurement uncertainty ranged from 2.01% to 34.4%. Method-matched calibration curves were used for the calculation of LOD and LOQ. The ion ratios of the qualifier and the quantifier ions were also monitored to successfully qualify the EU criteria to confirm the positive identification of a compound. The validation results of the current multi-mycotoxin method matched with the required European Commission Performance Criteria [49].

Mycotoxin Levels
The validated method was further used to investigate the incidence and content of mycotoxins in the collected rice samples. The results of the occurrence and levels of mycotoxins in rice samples are summarized in Table 2. Here, the mean is the average of mycotoxin contamination levels in all the samples including the negative samples, while considering their level as zero. The frequency of contamination of samples with AFB 2 and AFB 1 was 48% and 56%, respectively, within the concentration range of 1.5-9.15 µg/kg and 1.5-40 µg/kg, respectively. AFG 1 and AFG 2 were not detected in the tested samples. The results indicated that FB 1 was found in 42% (level range from 26 to 75 µg/kg) and ZEN in 15% (level range from 13 to 114 µg/kg) of the collected samples. OTA was detected in 6% of the samples within a concentration range of 2.5-11 µg/kg. Trichothecenes including DON, NIV, HT2, and DAS occurred in 8%, 28%, 10%, and 23% samples, and the mean contamination level of these mycotoxins were 6.99, 13.9, 2.03, and 1.60 µg/kg, respectively.  Figure 1 shows the mean contamination levels in the positive samples only (above the LOQ). The incidence rate of the mycotoxins was higher in North Punjab than in South Punjab. The levels of contamination of mycotoxins in both regions were not significantly different (p < 0.05). Overall, 78% of the samples were contaminated with either one or more mycotoxins, while 57% of the samples were contaminated with more than one mycotoxins. In 40% of the samples, AFB 1 and AFB 2 co-occurred. In 21% of the samples, the co-occurrence of the mycotoxins produced from Fusarium and Aspergillus moulds was observed. The co-occurrence of these specific toxins can be taken into account in the design of future toxicity studies to examine their impact on animal and human health.  Figure 1 shows the mean contamination levels in the positive samples only (above the LOQ). The incidence rate of the mycotoxins was higher in North Punjab than in South Punjab. The levels of contamination of mycotoxins in both regions were not significantly different (p < 0.05). Overall, 78% of the samples were contaminated with either one or more mycotoxins, while 57% of the samples were contaminated with more than one mycotoxins. In 40% of the samples, AFB1 and AFB2 cooccurred. In 21% of the samples, the co-occurrence of the mycotoxins produced from Fusarium and Aspergillus moulds was observed. The co-occurrence of these specific toxins can be taken into account in the design of future toxicity studies to examine their impact on animal and human health.

Consumption of Rice
On the basis of this survey, the mean rice consumption in the adult population (16-65 years) at national level was 108 ± 24.1 g per individual on a daily basis, while the rice intake (g) per kg of body weight (b.w.) was 1.92 ± 0.624 g (Table 3). Similarly, the mean rice consumption per capita and daily intake of rice per kg of body weight in children (7-15 years) at national level were estimated to be 56.65 ± 15.3 g/individual/day and 1.97 ± 0.619 g/kg b.w./day, respectively. The difference in rice consumption between adults and children of both regions was not significant (p < 0.05). In all the cases, the mean at national level was also not significantly different.

Consumption of Rice
On the basis of this survey, the mean rice consumption in the adult population (16-65 years) at national level was 108 ± 24.1 g per individual on a daily basis, while the rice intake (g) per kg of body Toxins 2018, 10, 77 6 of 30 weight (b.w.) was 1.92 ± 0.624 g (Table 3). Similarly, the mean rice consumption per capita and daily intake of rice per kg of body weight in children (7-15 years) at national level were estimated to be 56.65 ± 15.3 g/individual/day and 1.97 ± 0.619 g/kg b.w./day, respectively. The difference in rice consumption between adults and children of both regions was not significant (p < 0.05). In all the cases, the mean at national level was also not significantly different. Table 3. Consumption data (mean ± SD g/individual/day) and intake (mean ± SD g/kg b.w. per day) of rice by children and adults in different regions of Pakistan and at national level (mean) obtained during the survey in 2015-2016.

Effect of Cooking on AFB 1 and AFB 2 Levels
During cooking, the first step involved washing that caused a 15% reduction of aflatoxins. The reduction in AFB 1 (41-51%) and AFB 2 (47-63%) in three cooking recipes of rice was measured (Table 4). Aflatoxin reduction was significantly higher (p < 0.05) in Biryani than in other cooking recipes. AFB 1 reduction after cooking Pulao was in parallel with the reduction in boiled rice. The reduction level was different among these three cooking recipes because of differences in the cooking method. In boiled rice, the excess water was kept without adding any spices. In contrast, in Biryani recipe, the excess water was drained out after boiling the rice, spices (Coriander seed, cloves, cinnamon, ginger, black pepper, cardamom black, dry curry leaves, chili, and cumin seed) were added, and the rice was further cooked.

Exposure Assessment of Mycotoxins in Pakistani Children and Adults
The deterministic exposure levels associated with the mycotoxins AFB 2 , AFB 1 , NIV, DAS, FB 1 , OTA, DON, HT2 toxin, and ZEN are shown in Tables 5 and 6. The dietary exposure levels of AFB 1 and AFB 2 were taken after considering the mean reduction by cooking. We considered the two approaches, the mean concentration approach using fixed mycotoxin concentrations and variable values (mean, median, maximum, and percentiles) of consumption, and the second approach using a fixed consumption level with variable values (mean, median, maximum, and percentile) of  Tables 5 and 6, while the results of the other two scenarios are given as supplementary material (Tables A2-A5).
In this study, the estimated AFB 1 and AFB 2 intake in Pakistani adults and children through rice consumption was above the safe limits at all calculated exposure levels. The exposure of OTA for both adults and children of SP and of NP exceeded the PMTDI of 17.1 ng/kg b.w./day only at the maximum level [21]. There was no health risk associated with NIV exposure at the measured contamination levels as none of the exposure levels exceeded the safe health limits for NIV (1200 ng/kg b.w./day) [17].
All the determined exposure levels of HT2-toxin were found lower than the PMTDI value of 100 ng/kg b.w./day [18]. The exposure levels of DAS were compared with the TDI values of HT2 and T2 because of their structural similarity. All the estimated exposure levels of DAS, including the maximum exposure, were lower than the PMTDI value of HT2 and T2 (100 ng/kg b.w./day) [17]. However, the consumers' maximum exposure level for ZEN in NP showed a risk, as the values were higher than the TDI value (250 ng/kg b.w. day) [19].
Finally, when the two groups of subjects (adults and children) were compared, the children were found to be a vulnerable group for most contaminants, as the exposure was expressed in terms of per kg of body weight. The study reports, for the first time, the multiple-mycotoxins prevalence in Pakistani rice in combination with a consumption pattern of rice throughout Punjab, Pakistan, and the degradation of aflatoxins by the local cooking methods. This study also provides the first insight on the exposure to multiple mycotoxins associated with rice intake in Pakistan. The exposure of rice consumers in Pakistan (adults and children) to aflatoxins is above the toxicological reference values of AFs at all consumption levels. A combined intake of different mycotoxins at different concentration levels may lead to a higher risk than their single intake [50]. Nonetheless, we can conclude that the dietary exposures associated with the consumption of rice are considered a risk for public health. Hence, management strategies need to focus more on the reduction of mycotoxin contamination.

Cancer Risk and Margin of Exposure
The risk characterization due to AFB 1 exposure by rice consumption was measured using the margin of exposure (Table 7) and the liver cancer risk approach ( Table 8). The margin of exposure was calculated using Bench Mark Dose Level for 10% increased cancer risk based on the rodent data (BMDL 10 ) which is 170 ng/kg b.w./day [51]. An MoE greater than 10,000 is considered a low health concern [23]. The MoE values were calculated at exposures of fixed AFB 1 concentration and variable rice consumption by adults and children of SP and NP at lower bound (LB) and upper bound (UB) scenarios (Table 7). According to the results, the population in NP is at a higher risk compared to that of SP. In addition, all the MoE values were <10,000 in both regions, indicating a high risk due to AFB 1 exposure through rice consumption in Pakistan. AFB 1 , a potent human carcinogen, is synergistically linked to human primary liver cancer, highlighting a cancer burden in developing countries. The risk of liver cancer in individuals exposed to chronic hepatitis B virus (HBV) infection together with aflatoxins is 30 times higher than the risk in individuals exposed to aflatoxins only [52]. The age standardized rate (ASR) of liver cancer (HCC) is 3.6 cases per 100,000 individuals per year in the Pakistani population, with a 2.9% incidence [53]. The weighted average of hepatitis B antigen prevalence was 2.4% (range 1.7-5.5%) in children and 2.4% (range 1.4-11%) in the adult population [54]. Taking the HBV prevalence rate in Pakistan into account, the risk estimates of aflatoxin-induced HCC cases were calculated by the deterministic approach using a fixed mean AFB 1 concentration and a variable consumption.  The calculated mean cancer risk in adult and child populations of SP in Pakistan were 0.070 and 0.071 cases per year per 10 5 individuals, respectively, whereas the risk of development of liver cancer in adults and children of NP were calculated to be 0.122 and 0.127 cancers/year/10 5 /ng AFB 1 /kg b.w./day, respectively. The results showed that, at maximum, the total liver cancer risk associated with rice consumption in Pakistan was 0.725 and 0.752 HCC cases/year/100,000 individuals in adults and children, respectively.

Mycotoxin Contamination
The natural occurrence of multi-mycotoxins in rice has been studied in many regions of the world [32,33]. The present work is the first study on a multi-mycotoxin survey in rice in Pakistan. This work investigates the incidence of the major Aspergillus and Fusarium mycotoxins, as well as of the minor Alternaria mycotoxins in rice by using LC-MS/MS. In previous studies on Pakistani rice [24][25][26][27][28][29][30][31], information about only the traditional mycotoxins, such as aflatoxins and ochratoxin A, was reported. The occurrence of AFB 1 and OTA reported in Pakistani rice was more or less similar [26,29]. A previous study on Pakistani rice (n = 62) recorded that 47% and 37% of samples contained AFB 1 and OTA in a concentration range of 0.04-21.3 and 0.6-25.4 µg/kg, respectively [26]. Furthermore, our earlier study demonstrated that 56 and 50% of rice samples were positive for aflatoxins and OTA, with levels of contamination ranging from 0.05 to 24 and 0.06 to 15 µg/kg, respectively [29].
Studies on multi-mycotoxin contamination of rice, however, have been conducted in some other countries [4,[55][56][57][58][59]. A high incidence of multiple mycotoxins was found in Nigerian, Turkish, Chinese, and Indian rice [4,56,58]. A study showed 100% frequency of positive samples (n = 21) for AFs in Nigerian rice. The levels of AFB 1 and AFB 2 were in the range of 4.1-309 µg/kg (37.2 µg/kg mean level) and 1.3-24.2 µg/kg (mean level of 8.3 µg/kg), respectively. OTA level was in the range of <LOD-341.3 µg/kg in 67% of the samples, with a mean level of 141.7 µg/kg, while ZEN was prevalent in 52% of the samples, with a mean level of 10.6 µg/kg (in the range of <LOD-41.9 µg/kg). The contamination level for DON was in the range of <LOD-112.2 µg/kg in 24% of the samples (mean 18.9 µg/kg). Moreover, 14% of the samples were positive for FB 1 , with concentrations in the range of 0.4-4.4 µg/kg and a mean of 0.2 µg/kg [58]. The levels of AFs and OTA contamination in Turkish rice were found to be in the range of 0.05 to 21.4 and 0.025 to 80.7 µg/kg, with a 58% and 72% incidence rate, respectively [56]. The data from an Indian study revealed the content of AFB 1 to be in the range of 0.1 to 308 µg/kg in 68% of rice samples (n = 1600) [4].
Some studies also reported lower contamination levels of mycotoxins in rice in different countries as compared to the present study [55,57]. A study on rice samples (n = 9) from Qatar, observed levels of AFs in the range of 0.14-0.24 µg/kg and of OTA in the range of 1.65-1.95 µg/kg. The level of DON found in one sample was 142.31 µg/kg and the level of ZEN was from 0.18 to 1.41 µg/kg [55]. A survey of rice samples (n = 199) sold in Canada revealed that 50%, 22%, and 8% of rice samples were contaminated with AFB 1 , OTA, and FB 1 , respectively. The concentration levels of AFB 1 , OTA, and FB 1 were lower as compared to the present study, ranging between 0.002-7.1 µg/kg, 0.05-0.49 µg/kg, and 0.7-14 µg/kg, respectively [57].
In the present study, a higher incidence of Aspergillus toxins was observed in comparison to Fusarium mycotoxins. However, in Korean brown rice, an opposite trend was reported, as the frequencies of AFs, ZEN, NIV, and DON were 7.5%, 84%, 29%, and 1.3%, respectively, whereas the levels of contamination were estimated from 0.7 to 2.7 µg/kg (AFs), from 4.2 to 201.3 µg/kg (ZEN), from 7.7 to 349 µg/kg (NIV), and 43.2 µg/kg (DON) [59]. The variation in mycotoxin contamination in rice from different regions may be due to differences in climate as well as differences in the prevalent toxigenic microflora influenced by different agricultural practices and storage conditions [33].

Consumption of Rice
Consumption of rice in Iran (107 g/day per person) [34], Japan (157 g/day per capita), and China (183 g/day per capita) are comparable to our consumption results but not to the very high rice consumption (377 g/head/day) in Thailand [60].

Effect of Cooking on AFB 1 and AFB 2 Levels
The reduction of aflatoxins during the cooking process might occur because of its removal during washing, binding to a food matrix, or degradation or modification to unknown products during the aqueous heating process. A study reported a 14.7-15.3% reduction of aflatoxins with washing similar to the results of the present study [61]. The reduction of aflatoxins during rice cooking was reported by previous studies [62][63][64][65]. Ordinary cooking of AFB 1 -contaminated rice gave an average reduction of 34% [64], while a 78-88% reduction was observed with pressure cooking [63]. In another study, cooking of AFB 1 -contaminated rice showed 87.5% (cooking in excess water), 84% (ordinary cooking), and 72.5% (microwave cooking) reduction of AFB 1 [62]. In contrast, a low level of AFB 1 reduction (24.8%) was observed with two cooking methods common in Iran [65].

Exposure Assessment of Mycotoxins in Pakistani Children and Adults
Average AFB 1 intakes of 22.2-22.3 ng/kg b.w./day [26] and 19.1-26.6 ng/kg b.w/ day [25] were reported in previous studies. Our study results showed lower exposure compared to these previous studies, whereas in the high consumption group the exposure values in our study (42.77 and 44.36 ng/kg b.w. day) were higher. This might be due to differences in sampling, levels of contamination of AFs in rice, and consumption data considered during these studies. By comparing the previous studies conducted in different countries, our results match with the estimated mean AFB 1 exposure due to rice consumption in Korea (0.89-5.37 ng/kg b.w./day) [41], Brazil (6.5-6.6 ng/kg b.w./day) [66], and Nigeria (5.2 ng/kg b.w./day) [37]. In contrast, in Japan, the observed dietary exposure of AFB 1 was 1.78 and 1.20 ng/kg b.w./day at the 95th percentile level in children aged 7-14 years and in adults, respectively [67].
Because of the low contamination levels, the dietary exposures to AFB 1 from rice and wheat intake in French adults (<LOD-0.018 ng/kg b.w./day) and children (<LOD-0.035 ng/kg b.w./day) [38], as well as in the Lebanese population (0.63-0.66 ng/kg b.w/ day) based on total diet intake were lower than those observe in the present study [68]. The current data is higher than the reported mean dietary exposure to AFB 1 (0.8 and 0.12 ng/kg b.w./day) by rice consumption in Thailand in the years 2012 and 2013 [40]. Also, the dietary exposure to AFB 1 (0.033 ng/kg b.w./day) in Morocco [44] and that to total AFs in French adults (0.014 ng/kg b.w./day) and children (0.077 ng/kg b.w./day) [39] are lower than those observed in the current study. On the other hand, the mean AFB 1 exposure values of 21.7 ng/kg b.w./day in Northern Vietnam adults, 33.7 ng/kg b.w./day in children [43], 22.2 ng/kg b.w./day in adults of Lao Cai province of Vietnam [42], 5.8-76 ng/kg b.w./day (LB-UB) in the Chinese population [69], and 14 ng/kg b.w./day in the African country Gambia were higher than those observed in our study, because of high rice consumption and contamination levels [45]. The mean dietary exposures of 1.50 ng/kg b.w./day of OTA in the Nigerian population [37] and of 4.28 ng/kg b.w./day in the Lebanese population [68] were comparable to thos eobserved in this study. The higher mean dietary exposures to OTA (24.2-24.7 ng/kg b.w./day) by rice consumption reported in Pakistan [26] and in Vietnam (7.9 ng/kg b.w./day) were due to relatively higher contamination levels of OTA in rice [42]. The mean dietary exposure of OTA was higher in Chinese adults (4.62 ng/kg b.w./day) and children (13.9 ng/kg b.w./day) because of a higher consumption level of cereals [70]. The mean OTA exposure values of 0.0004 ng/kg b.w./day in US adults and 0.001 ng/kg b.w./day in US children aged >6-18 years [35], 0.02 ng/kg b.w./day in Turkish rice consumers [36], 0.17 ng/kg b.w./day in Spain [46], 0.08 ng/kg b.w./day in French adults, 0.19 ng/kg b.w./day in French children [40], and 0.0428 ng/kg b.w./day in Moroccans [44] by rice consumption were lower than those reported in our study.
Similar results were found for DON mean dietary exposure of 9.51 ng/kg b.w./day in French adults and 23.7 ng/kg b.w./day in children by rice intake [39], as well as 5.7-6.6 ng/kg b.w./day in French adults and 14.3-16.0 ng/kg b.w./day in children by rice and wheat intake [38]. Through cereal intake, a higher dietary exposure of DON was reported for Chinese adults (309 ng/kg b.w./day) and children (927 ng/kg b.w./day) compared to what found in the present study [70]. The estimated mean exposure values for DON, 5.97 ng/kg b.w./day in the Nigerian population [37] and 0.18 ng/kg b.w./day in the Moroccan population [44], and 1.56 ng/kg b.w./day in the Lebanese population [68] were lower than those reported in the current study data. The mean estimated exposure values of 3.8 ng/kg b.w./day in French adults, 42.5 ng/kg b.w/ day in children [40], 0.042 ng/kg b.w./day in the Moroccan population [45], and 19.13 ng/kg b.w./day in the Nigerian population [37] for FBs by rice intake were lower than those found in the present study. In contrast, a higher FBs mean exposure of 536 ng/kg b.w./day in Northern Vietnam adults and 1019 ng/kg b.w./day in children was reported [43].
In the present study, NIV exposure results are similar to those reported for the Nigerian population (20.81 ng/kg b.w./day) [37]. The present study showed a higher level of the estimated mean exposure for NIV by rice intake as compared to that reported in French adults (4.86 ng/kg b.w./day) and children (11.2 ng/kg b.w./day) [39] and in the Moroccan population (6.12 ng/kg b.w./day) [44], as well as to the mean exposure ranges (UB-LB) of NIV in French adults (1.88-2.8 ng/kg b.w./day) and children (4.76-6.43 ng/kg b.w./day) [39]. Data on the exposure to HT-2 and DAS is scant in the literature. In this study, the HT2 toxin dietary intakes were higher than the mean estimated exposure for the sum of HT2 and T2 (0.361 ng/kg b.w./day) in the Moroccan population [44] and the mean exposure ranges (UB-LB) for HT2 in French adults (0.19-1.53 ng/kg b.w./day) and children (0.16-1.42 ng/kg b.w./day) [38]. The mean estimated exposure (0.101 ng/kg b.w./day) of DAS by rice intake in the Moroccan population [44] was lower as compared to our values.
The mean dietary exposure values of 2 ng/kg b.w./day for ZEN by rice consumption in Iranian people [34] and in French adults (1.63 ng/kg b.w./day) and children (3.75 ng/kg b.w./day) were lower as compared to those found in our study [39]. Also, the mean estimated exposure ranges (UB-LB) for ZEN by rice and wheat products intake in French adults (0.06-0.7 ng/kg b.w./day) and children (0.16-1.42 ng/kg b.w./day) were lower than those observed in the present study [38]. A higher estimated mean exposure of 157.36 ng/kg b.w./day of ZEN was reported in the Nigerian population [37] by rice intake, and of 155 ng/kg b.w./day in Chinese adults and of 464 ng/kg b.w./day in Chinese children through cereal intake [70].
These discrepancies in exposure estimations are due to the variability in food consumption levels, differences in contamination data, cooking effects, and differences in the analytical techniques used in the reported studies.

Cancer Risk and Margin of Exposure
Based on these results, it is obvious that the intake of contaminated rice is of great public health concern. Our results for the estimated cancer risk and MoE due to AFB 1 intake by rice consumption are comparable with those of previous studies reported for Japan [67] and Thailand [40]. The study for Japan measured a cancer risk of 0.031 cancer/year/10 5 , and a MoE of 141 for children (age 7-14 years) resulting from AFB 1 intake by rice consumption, while in adults the values of cancer risk and margin of exposure were 0.021 cancer/year/10 5 and 209 MoE, respectively [67]. The mean cancer risk and margin of exposure due to AFB 1 intake through colored and brown rice in all age groups of Thailand were in the range of 0.010-0.039 cancer/year/10 5 , with 5% HBV prevalence rate [40].
The mean cancer risk and margin of exposure due to total dietary AFB 1 intake in Brazil [66] and Malaysia [71] were also similar to those determined in our study. The mean cancer risk and margin of exposure in Brazil, with a 0.37% prevalence rate of HBV, were 0.0753 cancer/year/10 5 and 25 MoE, respectively [66]. The liver cancer risk in Malaysia by total dietary AFB 1 calculated by eliminating high contamination data was 0.01-0.26 cancer/100,000 people/year, with a 0.2-2.1% contribution to liver cancer cases, while, by adding the high contamination data, the value increased to 0.61-0.85 cancer/year 10 5 people, raising the contribution to liver cancer cases up to 12.4-17.3% [71].
The cancer risk and margin of exposure due to AFB 1 intake through rice in Chinese [69], Vietnamese [42], and African populations [45] were higher as compared to those found in the present study. This might be because of high contamination levels of AFB 1 in rice and high consumption levels, as well as high prevalence rate of HBV in those countries. The cancer risk and margin of exposure in the Chinese population were 0.2-2.65 (UB-LB) cancer/year/10 5 people and 1.8-24.1 (UB-LB) MoE, respectively, using a 14.3% prevalence rate of HBV [69]. The mean cancer risk and margin of exposure in Vietnam were 1.51 cancer/year/10 5 individuals and 8 MoE, respectively, using a 20% prevalence rate of HBV [42]. Another study in Africa reported a cancer risk of 1.1 cancer/year/10 5 people considering a 25% HBV prevalence rate and an MoE of 12.1 for Gambia [45]. The estimated mean cancer risk of 2.3 cancer/year/10 5 people for the consumers was reported for Northern Vietnam, considering a 20% HBV prevalence rate, while, in adults, the value of cancer risk was 1.2 cancer/year/10 5 people [43]. Because of the low contamination level in foods, the estimated dietary cancer risk in Japan is much lower as compared to our study results: even at the 99.9th percentile, it was 0.00059-0.00067 cancer/year/10 5 population assuming a 1% HBV prevalence rate [72].
To reduce the synergistic effect of HBV and aflatoxins, the improvement in HBV vaccination as well as measures for the reduction of aflatoxin contamination in foods are necessary. The risk characterization of multiple-mycotoxins in rice will be useful for setting mycotoxin priority in adopting control measures. Moreover, the measured mycotoxin exposure due to rice in this study can be used in future total dietary exposure estimates in Pakistan, as the burden of cancer risk depends on the cumulative exposure. In addition to HBV and aflatoxins, Hepatitic C Virus (HCV) may also be taken into account in future coexposure studies, since it is one of the leading factors involved in HCC. Notably, HBV [54] and HCV [73] are highly prevalent in the Pakistani population.

Conclusions
The study measured the levels of multi-mycotoxin contamination in rice samples using a validated multi-mycotoxin LC-MS/MS method. The risk assessment of mycotoxins by rice intake in adults and children of the Pakistani population in SP and NP regions was performed after attaining consumption data of rice by conducting food frequency questionnaires. The mycotoxin contamination profile of Pakistani rice showed the prevalence of AFB 2 , AFB 1 , NIV, DAS, FB 1 , OTA, DON, HT2, and ZEN. Among the evaluated rice cooking recipes, Biryani demonstrated the highest degradation of AFB 2 (63%) and AFB 1 (51%). Our results indicate that dietary exposure to mycotoxins through rice was higher in children as compared to adults, and higher in NP in comparison to SP. The MoE of AFB 1 was remarkably lower than the recommended safe limits. Moreover, there is a potential risk, due to rice consumption, of developing aflatoxin-induced hepatocellular carcinoma (HCC) in Pakistan. The Pakistani population is not exposed to a single mycotoxin but faces exposure to multiple mycotoxins. This study highlights the need to establish regulatory guidelines regarding the prevalence of mycotoxins in Pakistani foods, and a regular monitoring of highly consumed foodstuff, especially rice and cereals, is suggested. A cumulative risk assessment from the exposures of multi-mycotoxins, especially in the HBV-and HCV-infected population, also needs to be thoroughly studied in the future.

Sampling and Food Consumption Data
Polished rice of all varieties and brands, intended for human consumption, was randomly purchased in the quantity of 1 kg from different wholesale markets, super markets, and small shops located in ten districts of two different agroecological zones of Punjab (North and South Punjab), Pakistan, during the year 2015. Ninety samples from each region were collected ( Figure 2). The average rainfall in South Punjab (SP) and North Punjab (NP) are between 22.65 mm and 66.99 mm, respectively. The average annual temperatures in SP and NP are 26 • C and 24 • C, respectively. The location of SP and NP in coordinates (latitude and longitude) are between 28-30 • N, 70-71 • E, and 31-33 • N, 72-74 • E, respectively [74,75].
All samples were ground using an M 20-grinder (Ika-Werke, Staufen, Germany) and kept in plastic bags at −20 • C before mycotoxin determination. To obtain accurate exposure estimates, rice consumption data were obtained by conducting a survey in southern and northern Punjab regions of Pakistan. A FFQ was prepared, and individuals and families were interviewed. Portion-size pictures (small, medium, and large servings of cooked rice, i.e., 50 g, 75 g, 100 g of uncooked rice) were used to gather information on the rice intake, and the actual weight of each portion size was measured. The diet intake information for one week was gathered from the participants, and the mean daily rice intake of each individual was calculated (per kg of body weight per day). The proportions of participants from South Punjab (SP) and North Punjab (NP) regions in gathering the consumption data were 48% and 52%, respectively. The gender distribution of the participants was 48% male and 52% female. In total, 548 individuals in the adult category were interviewed, and the data of 467 individuals in the children category (age 7-15 years) was gathered by interviewing either the children or the female family head. Finally, the data from both regions was arranged separately for each category (adults and children) to get the mean, median, minimum, maximum, and percentile (P75, P90, P95) intake of rice. Furthermore, the generated consumption data was used in calculating the dietary exposure to mycotoxins.

Sampling and Food Consumption Data
Polished rice of all varieties and brands, intended for human consumption, was randomly purchased in the quantity of 1 kg from different wholesale markets, super markets, and small shops located in ten districts of two different agroecological zones of Punjab (North and South Punjab), Pakistan, during the year 2015. Ninety samples from each region were collected (Figure 2). The average rainfall in South Punjab (SP) and North Punjab (NP) are between 22.65 mm and 66.99 mm, respectively. The average annual temperatures in SP and NP are 26 °C and 24 °C, respectively. The location of SP and NP in coordinates (latitude and longitude) are between 28-30° N, 70-71° E, and 31-33° N, 72-74° E, respectively [74,75].

Sample Preparation
The sample extraction methodology described by Monbaliu et al. [76] was followed. Internal standards were added to the samples before extraction. Five grams of the rice sample was extracted with 20 mL of acetonitrile/water/acetic acid (79/20/1, v/v/v) by agitating on a vertical shaker for 1 h and centrifuged for 15 min at 3300 g. The supernatant was subjected to cleanup by octadecyl (C 18 ) solid phase extraction (SPE) column (Grace octadecyl C18, Lokeren, Belgium) on a vacuum elution manifold after conditioning with 10 mL of the extraction solvent-acetonitrile/water/acetic acid (79/20/1, v/v/v). The extraction was performed a second time by adding 5 mL of the extraction solvent to the samples. The eluate was collected in a 25 mL volumetric flask. The volume was adjusted with the extraction solvent. The extract was defatted with 10 mL n-hexane. Then the extract was split into two parts to perform two different modes of cleanup. In the first cleanup, 12.5 mL of the defatted extract was diluted with 27.5 mL of acetonitrile/acetic acid (99/1, v/v), and 30 mL of this extract was passed through Multisep226, Afla-ZON + Multifunctional columns from Romers Lab. (Gernsheim, Germany), followed by washing with 5 mL of acetonitrile/acetic acid (99/1, v/v). In the second cleanup mode, the defatted extract (10 mL) was filtered through a Whatman glass microfilter (VWR International, Zaventem, Belgium), and 2 mL of this filtered extract was combined with the MultiSep 226 eluate.
The combined eluates were evaporated, and the residue was dissolved in 150 µL of mobile phase containing methanol/water/acetic acid (57.2/41.8/1, v/v/v) and 5 mM ammonium acetate. Before LC-MS/MS analysis, the resulting solution was ultracentrifuged for 5 min at 14,000 g using ultra free-MC centrifugal filters (Bedford, MA, USA).

Analysis by LC-MS/MS
A micromass Quatro Micro triple quadrupole mass spectrometer (Waters, Milford, MA, USA) equipped with a Waters Acquity UPLC system was used to analyze the samples, and the Masslynx (4.1) software (Micromass, Manchester, UK) was used for data processing. The analytical column was a Symmetry C18, 5 µm, 2.1 × 150 mm (Waters, Zellik, Belgium), with a guard column of Waters Sentry 3.5 µm 2.1 × 10 mm (Waters, Zellik, Belgium). The column and autosampler temperature was kept at 30 • C, and 20 µL was injected. Capillary voltage was set at 3.2 kV with a source voltage of 150 • C and a 350 • C desolvation temperature. Liquid chromatography conditions (mobile phase composition and gradient) and MS parameters were followed as described by Monbaliu et al. [77].

Quality Control and Quality Assurance
A set of performance characteristics that were in compliance with the recommendations defined by EU Commission Regulation EC/401/2006 was evaluated [49]. Deepoxy-deoxynivalenol (DOM) and zearalanone (ZAN), structural analogues of the type-B trichothecenes, and ZEN were used as internal standards in the multi-mycotoxin determination to compensate for matrix effects and for losses during extraction and cleanup. For each mycotoxin, a spiking experiment was performed at five concentration levels except for aflatoxins, which were spiked at six different levels each day in triplicate for four validation days (Table A1). The validation parameters assessed were: linearity, apparent recovery, limit of detection (LOD), limit of quantification (LOQ), intraday repeatability (RSD r ), interday repeatability (RSD R ), and expanded uncertainty. The linearity was tested graphically using a scatter plot, and the linear regression model was evaluated using a lack-of-fit test. The apparent recovery was calculated by dividing the observed value (quantified using calibration plot) by the spiked level. The sensitivity of the method was estimated by LOD. A series of blank rice samples spiked at low concentration levels were used to estimate LODs and LOQs, which provided a signal to noise ratio of 3:1 and 10:1 for the weakest transitions in LC-MS/MS chromatograms for each of the analyte, respectively. The precision in terms of intraday repeatability (the analysis of three replicates on the same day) and interday repeatability (the analysis of three replicates on four different days) was calculated using relative the standard deviation (RSD) at the spiked concentration levels (n = 5). The expanded measurement uncertainty (U) was obtained by multiplying the combined standard uncertainty (uc) by a coverage factor k = 2, based on the desired level of confidence of approximately 95%, where the uc was an estimated standard deviation calculated as the positive square root of the total variance obtained by combining the intralaboratory repeatability (sR), the uncertainty associated with the purity of the standards (U (Cref)), and the uncertainty associated with the mean recovery (sbias).

Cooking of Rice by Pakistani Recipes
Three most common local cooking methods of rice in Pakistan were evaluated for their efficiency in the degradation of AFB 1 and AFB 2 . Negative control samples were also washed and cooked as in each treatment, and spiked with known concentrations of standards after cooking. The aim was to make matrix-matched calibration curves and quantify the levels of AFB 1 and AFB 2 in the treated samples. The positive controls were samples of naturally contaminated uncooked rice that were used for comparison of the treatments to conclude the percentage of AFs degradation.
Naturally contaminated rice samples (100 g each in triplicate for each treatment) were washed with water (200 mL) three times, soaked in 200 mL of water for 20 min, and finally the water was completely removed. This washing step was similar in all treatments. For the first recipe (treatment) of boiled rice, the washed rice was added to boiling water (200 mL) and cooked for about 20 min. In the second recipe (pulao), first a curry (ingredients: oil, onion, ginger and garlic paste, tomato, boiled chicken, and salt and chili) was prepared, and 200 mL of water was added to the curry. On boiling, the washed rice was added to the mixture and cooked for 10 min at high flame and 20 min at low flame, while covering it tightly with a lid. For the third recipe (Biryani), the washed rice was boiled in excess water for 5 min, and the water was removed. The curry was prepared separately, having the same ingredients as those in pulao with additional yogurt and spices like cumin, pepper, cloves, cinnamon, cardamom, bay leaves, coriander, and mint leaves. Layers of curry and the boiled rice, alternatively on top of each other, were made in a pot and further cooked for 20 min at low flame after covering tightly with a lid. The cooked rice was cooled down and freeze dried. Then the samples were ground and analyzed for aflatoxin levels after sample preparation, following the methodology described by Majeed et al. [78].

Dietary Exposure Assessment
The dietary exposure of mycotoxins was calculated by a deterministic risk analysis (Equation (1)).
Dietary Exposure = concentration of mycotoxin × daily rice intake per kg of body weight (1) The left-censored mycotoxin contamination data related to the non-detects (ND), and those below the limit of quantification (<LOQ) can be a source of uncertainty in exposure models [79]. So, three different scenarios (lower bound, medium bound, and upper bound) were incorporated in this study to cope with the uncertainty, following the approach described by EFSA [80]. The dietary exposure levels were estimated considering two approaches, using a fixed mycotoxin concentration and variable values (mean, median, maximum, and probability values) of consumption level, and using fixed consumption levels with variable values (mean, median, maximum, and probability values) of mycotoxin levels.

Risk Characterization
The risk characterization of the genotoxic aflatoxins was performed by both margin of exposure MoE [23] and cancer risk approaches [22]. The MoE was estimated (Equation (2)) by the ratio of Bench Mark Dose Level (BMDL) that causes a 10% increase in the cancer incidence in rodents (BMDL 10 = 170 ng/kg b.w. day) and the exposure to AFB 1 [23].
Margin of exposure = BMDL 10 /exposure (ng/kg b.w. day) The risk of AFB 1 -induced cancer (hepatocellular carcinoma, HCC) was calculated (Equation (3)) by multiplying the probability of cancer with the AFB 1 exposure estimates of min, max, mean, and percentiles in both upper and lower bound scenarios for each category in both regions. Here, cancer potency P cancer (Equation (4)) deals with the percentage of both carriers (%Pop.HBsAg + = 0.024) and noncarriers (%Pop.HBsAg − = 0.976) of HBV infection in the Pakistani population, that is 2.4% [54], as well as with the carcinogenic potency of AFB 1 for carriers (P HBsAg+ = 0.3 cancer/year/10 5 individuals) and noncarriers (P HBsAg-= 0.01 cancer/year/10 5 individuals).

Statistical Analysis
The normality of the consumption data distribution and contamination data was assessed by Kolmogorov-Smirnov, Shapiro-Wilk test, and the corresponding Q/Q plots. A non-parametric Mann-Whitney U test was applied to determine the significance, using the SPSS statistical package (IBM ® , Version 14, SPSS Inc. Chicago, IL, USA, 2005) with a level of confidence of 0.05. All other calculations were executed in Excel 2010. The mean data together with standard deviations (SD) was stated. Significant differences in percentage degradation in AFB 1 and AFB 2 among various cooking processes were determined by Tukey's HSD test (IBM ® , Version 14, SPSS Inc. Chicago, IL, USA, 2005).