Blood Mercury Levels in Relation to Overweight in Korean Children and Adolescents: Results From The Korean National Health and Nutrition Examination Survey 2010-2013

Background: Previous researches for adults have demonstrated a substantial role of mercury exposure in the development of overweight and obesity development, but those ndings are inconsistent. Although children and adolescents are more susceptible to the toxic effect of mercury compared to adults, the research of overweight and obesity related to mercury exposure in children and adolescent is scarce. This study aimed to investigate the association of blood mercury levels with body mass index (BMI) in Korean children and adolescents. Methods: The cross-sectional data analyzed were obtained from 1327 participants (age: 10-18 years, 672 males and 655 females) who completed the Korean National Health and Nutrition Examination Survey 2010-2013. The adjusted covariates were age, sex, household income, total intake of seafood in the previous 24 hours, and current status of alcohol consumption and smoking. Results: The adjusted geometric mean blood mercury level was 2.19 µg/L and increased signicantly in the overweight group and the highest seafood consumption quartile compared with the level in the normal weight group and lowest seafood consumption quartile, respectively (P < 0.001). The adjusted blood mercury levels were signicantly positively correlated with BMI in all participants and females (P < 0.05). When stratied by age and sex, all participants and males in the highest blood mercury level quartile were at higher risk of overweight than those in the lowest quartile (odds ratios [95% condential intervals, CIs]: 1.78 [1.06-2.98], and 2.06 [1.01-4.23], respectively) after adjusting for the covariates (P < 0.05). Receiver operating characteristic analysis of blood mercury levels for predicting overweight revealed that the area under the curve was 0.603 (P < 0.001). Conclusions: Our study suggested a signicant association between mercury exposure and the risk of overweight in Korean children and adolescents. with elevated blood mercury levels; however, research on the relationship of BMI with blood mercury status in children and adolescents is insucient. this study to investigate the association of blood mercury levels with BMI in Korean children and adolescents using data from the Korean National Health and Nutrition Survey (KNHANES) 2010-2013.

The KNHANES is a series of nationally representative population-based cross-sectional surveys about health and nutritional status that have been conducted using a strati ed, multistage, and probability sampling design by the Korea Centers for Disease Control and Prevention (CDC) [17]. Our study used data from the questionnaire, physical examination components, laboratory indices, and diet assessment of the KNHANES V 1-3 and VI-1 (2010-2013). Among an initial enrolled sample of 15,261 men and 18,291 women, we excluded those who were aged younger than 10 years and older than 19 years; had no data on anthropometric measures; had no responses on the questionnaires for household income, current status of alcohol consumption and smoking, and total amount of seafood consumed; and had no measurement of blood mercury levels or hematocrit (Hct) (Figure 1). The nal number of participants was 1,327 (10-18 years of age, 672 males and 655 females), and the participants were divided into normal and overweight groups. Overweight was de ned as a BMI ≥ 85 th gender and age-speci c percentiles based on the 2017 Korean growth chart [18]. The rest of the participants composed the normal group. BMI was de ned as the participants' weight in kilograms divided by the height in meters squared (kg/m 2 ). Participants were additionally divided into male and female groups.

Measurements of Blood Mercury Levels in Whole Blood
To measure the levels of mercury in whole blood, 3-mL blood samples were obtained from the participants after an 8-hour overnight fast. The blood mercury levels were measured using the gold amalgam collection method with a Direct Mercury Analyzer 80 (DMA-80, Milestone, Bergamo, Italy) in the Neodin Medical Institute (Seoul, South Korea), a central laboratory certi ed by the Korean Ministry for Health and Welfare [19]. For internal quality assurance and control, commercial reference material was used (Lyphochek ® Whole Blood Metals Control; Bio-Rad, Hercules, CA, USA) with coe cients of variation of 1.59-4.86% among four reference samples. For external quality assurance and control, the Neodin Medical Institute was approved by the German External Quality Assessment at Friedrich-Alexander University and by the Quality Assurance Program at the Korea Occupational Safety and Health Agency. The limit of detection for blood mercury using this method was 0.158 μg/L, and no participants presented a blood mercury concentration below this level in this study. Blood mercury levels were categorized into quartiles (Qs) and strati ed by sex and age. Q4 of blood mercury levels was de ned as the highest value.

Determination of Covariates
Data were obtained from the questionnaires of sociodemographic, lifestyle and dietary factors from the KNHANES. Household income was determined according to the quartile values of equalized household income (total household income divided by the square root of the number of household members) [20]. The current status of alcohol consumption was classi ed as drinker or nondrinker, which was de ned according to consuming alcohol more than once in the last month. The current status of smoking was classi ed as smokers or nonsmokers, which was de ned according to smoking more than once in the last month. The total amount of sh, shell sh, and seaweed (seafood) consumed within the previous 24 hours was obtained using a 24-h dietary recall questionnaire administered by a trained nutritionist. Additionally, Hct was used as a covariate and categorized into quartiles. For all quartile categorizations, Q4 is de ned as the highest value for each variable.

Statistical Analysis
Statistical analyses were performed using SPSS statistical software (version 22; IBM Co., Chicago, IL, USA) and visualized by R program (version 3.6.2). Logarithmic transformation was applied to skewed variables to improve the normality of the distribution. The complex weighted sample descriptive procedure was used for evaluating the numerical variables, and the complex weighted sample crosstabs procedure was used for evaluating the categorical or ordinal variables using weights provided by the KNHANES. The complex weighted samples general linear model was used to calculate the geometric means and 95% con dence intervals (CIs) to compare the blood mercury levels between the groups of study variables. The regression coe cient (95% CIs) of the association between blood mercury levels and BMI was calculated by the complex weighted samples general linear model after controlling the covariates, including the sociodemographic factors (sex, household income), dietary factors (seafood consumption), lifestyle factors (alcohol consumption, smoking), and blood Hct levels. Multivariate weighted logistic regression analysis was used to determine the odds ratios (ORs) and 95% CIs for predicting the risk of overweight, adjusted for the same covariates. Participants in Q1 of blood mercury levels were considered as the reference group. A receiver operating characteristic (ROC) curve was constructed to evaluate the performance of blood mercury levels in predicting overweight. The prevalence of overweight according to the quartile of blood mercury levels was calculated from the complex weighted sample crosstabs procedure. A two-tailed P value < 0.05 was considered statistically signi cant. P values for the trend test were performed in the overweight group across increasing blood mercury levels.
The characteristics of the participants are presented in Table 1. Among all participants, no signi cant differences between the normal and overweight groups in age, 3-year interval age groups, each survey year, blood Hct levels, household income quartiles, or the current status of drinking and smoking were found for males or females ( Table 1). The number of participants in the overweight group was 267 (20.1%), and in the normal group, the number was 1060 (79.9%). As expected, weight, BMI, and waist circumference were signi cantly higher in the overweight group than in the normal group among all participants, males and females (P < 0.05, Table 1). The overweight group showed signi cantly higher adjusted geometric mean blood mercury levels than the normal group among all participants, males, and females (P < 0.001, Table 1).
The overweight group showed signi cantly higher intake of seafood in the previous 24 hours than the normal group among females (P < 0.05, Table 1). Among all participants, the abdominal circumference in the male group was signi cantly higher than that in the female group (male: 71.62 ± 0.48, female: 67.75 ± 0.41, P = 0.000). The blood Hct levels in the male group were signi cantly higher than those in the female group (male: 43.44 ± 0.15, female: 39.89 ± 0.12, P = 0.000).
2. Blood Mercury Levels According to Participants' General Characteristics Table 2 presents the geometric means and 95% CIs of blood mercury levels according to participants' general characteristics. The geometric mean (95% CIs) of blood mercury levels in all participants, males, and females was 2.19 (2.06-2.26), 2.18 (1.98-2.38), and 2.14 (1.91-2.36) µg/L, respectively, after controlling for age, sex, total intake of seafood in the previous 24 hours, household income, and current status of drinking and smoking. No signi cant differences in crude or adjusted blood mercury levels were found between the male and female groups.
The geomean of blood mercury levels was signi cantly higher in the overweight group than in the normal group among all participants and the male group but not in the female group after adjusting for covariates (P < 0.01, Table 2). The Q3 and 4 of blood Hct levels showed signi cantly higher crude and adjusted geomean of blood mercury levels than the Q1 of blood Hct levels in all participants and the male group (P < 0.05, Table 2). Q4 of a total amount of seafood consumed in the previous 24 hours showed signi cantly higher crude and adjusted blood mercury levels than Q1 among all participants, males and females (P < 0.05, Table 2). No signi cant differences in blood mercury levels were found according to the current status of drinking and smoking ( Table 2).

Associations between Blood Mercury Level and BMI
Logarithmically transformed blood mercury levels were signi cantly positively correlated with logarithmically transformed BMI in unadjusted and adjusted models of all participants and females (P < 0.05, Table 3, Figure 2). In the male group, unadjusted and adjusted age and sex models showed signi cantly positive associations between logarithmically transformed blood mercury levels and logarithmically transformed BMI (P < 0.05, Table 3).

Prevalence of Overweight associated with the Quartile of Blood Mercury Levels
In all participants, males and females, the prevalence of overweight participants increased with the elevation of the blood mercury quartiles (P < 0.01, Table 5).

Discussion
In this study, we demonstrated that elevated blood mercury levels were associated with risks for overweight in Korean children and adolescents after adjusting for possible potential confounders according to data from the KNHANES.
Several studies in adults have shown that BMI was associated with blood mercury levels; however, they showed inconsistent results [5,6,9].
Moreover, few studies have been performed on children and adolescents, and these studies also showed inconsistent results [9]. One study showed that obesity is associated with blood mercury levels in children and adolescents, but this previous study had a limitation in regard to a lack of adjustment for seafood consumption, whereas our study performed this adjustment [21]. Another study in children revealed that BMI was inversely related to blood mercury levels, suggesting differences in mercury metabolism related to BMI between adults and children [9].
Our study revealed that blood mercury levels showed a signi cant positive association with BMI after adjusting for covariates, including seafood consumption. Additionally, participants in the highest quartile of blood mercury levels were at higher risk of overweight than those in the lowest quartile, and the blood mercury levels exhibited acceptable performance in predicting overweight in the ROC analysis. These results suggested that mercury exposure could result in an increased risk of overweight in Korean children and adolescents. Blood mercury levels could be related to age, pattern of seafood consumption, genetics involved in mercury clearance, and race [22]. Our study could provide additional valuable information on blood mercury levels associated with Korean overweight children and adolescents.
The most commonly accepted methods of assessing mercury exposure are to test urine or blood, which usually measure the level of total mercury (elemental, inorganic and organic) [23]. The blood mercury level re ects the burden of organic mercury that has accumulated in the body, whereas urinary mercury re ects exposure to inorganic mercury [23]. For most people, an elevated blood mercury level is associated with eating sh and other seafood containing organic mercury. In a study of a general Korean population, seafood intake during the previous 3 days before blood sampling affected blood mercury levels in both males and females [24]. In line with these results, our study showed that blood mercury levels were signi cantly higher in the highest seafood consumption quartile than in the lowest seafood consumption quartile in the previous 24 hours. The consumption of sh contaminated with a form of methylmercury is considered to be an important exposure pathway in children and adolescents.
In our study, blood mercury levels were signi cantly elevated in the overweight male group compared with the levels in the normal group after controlling for covariates, but not in the overweight female group. Possible explanations are that males have more visceral adipose tissue and less subcutaneous adipose tissue than females and that visceral adipose tissue contains higher toxin levels than subcutaneous adipose tissue [25][26][27]. Our study also showed higher waist circumferences in the overweight male group than in the overweight female group. However, no sex differences were demonstrated in the geometric mean value of blood mercury levels. This nding is consistent with the representative data of blood mercury levels in the Korean general population, which revealed sex differences in blood mercury levels in a population over 20 years old [24]. In rodent studies, sex differences in tissue distribution and clearance of mercury have been demonstrated, and the ndings suggest that these differences may be caused by androgens and estrogen [28,29]. Our result can be explained by the fact that sex hormones are not as present in participants aged 10-18 years.
Some studies of adults have reported that other factors associated with higher blood mercury levels are household income, alcohol consumption, and smoking [19,30]. In our study, blood mercury levels were not signi cantly associated with household income or the current status of alcohol consumption and smoking after adjustment for covariates. This inconsistency with previous reports could be related to the different de nitions of the status of smoking and alcohol consumption, different populations and small numbers of current alcohol consumers and smokers in our study. Hct is also considered to be a covariate related to blood mercury level because up to 80% of methylmercury binds to red blood cells [10]. Higher blood Hct levels are known to result in greater mercury binding in whole blood and thus a higher mercury concentration in the blood sample [31]. Our results showed that the blood Hct levels in males were higher than those in females, which is related to the report that the differences in hematologic variables between sexes emerge after the onset of menstruation [32]. After adjusting for these possible covariates related to blood mercury levels, our study suggested that blood mercury levels were a signi cant risk factor for overweight in Korean children and adolescents.
According to the KNHNES 2011-2013, the geometric mean blood mercury level in Korean adults was 3.37 g/L [5]. These mercury levels are approximately three to four times higher than those of adults in the U.S. (0.86 g/L), Germany (0.58 g/L) or Canada (0.76 g/L) [33][34][35]. A previous study in a general population noted that as the participants got older, the blood mercury levels increased [24]. Our study, conducted with children and adolescents aged 10-18 years, showed that the geometric means of the blood mercury levels were 2.19 g/L, 2.18 g/L, and 2.14 g/L in all participants, males and females, respectively, after adjusting for covariates. The mercury levels in our study were also much higher than the 0.8 g/L of the reference value in German children aged 3-14 years old and the 0.3 g/L of the median blood mercury level in American children and adolescents aged 1-17 years old [36,37]. Accordingly, the blood mercury level in Koreans is greatly higher than that in individuals in Western countries, including children and adolescents. This difference seems to be related to the fact that seafood is a staple food in Korea and can be a primary exposure pathway of methylmercury [38]. In our study, the mean intake of seafood was 54.18 g/day, which is higher than in Germany (37.5 g/day), Canada (24.7 g/day), and the U.S. (12.8 g/day) [39][40][41].
Many kinds of sh do not contain high levels of mercury [42]. While balancing concerns for methylmercury exposure, eating a variety of different types of sh is a good way to gain good nutrition because sh contain high-quality protein and omega-3 fatty acids. According to the Environmental Protection Agency (EPA)-FDA advice in the U.S., women of childbearing age (approximately 16-49 years old), pregnant and breastfeeding women, and young children should eat more sh that are low in mercury for important developmental and health bene ts [43].
They recommend that those people eat two to three servings (8-12 ounces for adults and children over age 10) of a variety of sh and shell sh each week. Mercury has been widely recognized as a threat to children and adolescents' health, and a global policy for seafood consumption is required to reduce and prevent mercury exposure.
This cross-sectional study has a limitation in that it is di cult to establish accurate cause-effect relationships between overweight and blood mercury levels. To our knowledge, no prospective studies have noted whether overweight/obesity develops or is aggravated after mercury exposure. To understand the underlying pathogenesis of this association between overweight and blood mercury levels, further studies are required. Additionally, since environmental conditions are changing, reference values are continuously checked and updated if new information becomes available. Recent national data for mercury exposure in the general population are required.
In conclusion, our study demonstrated an association of mercury exposure with the risk of overweight in Korean children and adolescents.
Further studies are needed to investigate the mechanism by which high blood mercury levels are associated with overweight and to establish global guidelines for seafood consumption to reduce mercury exposure in children and adolescents.  Values are presented as the mean ± standard error or number (%). P value by t-test or chi-square test. Abbreviations: BMI, body mass index;    Values are presented as the mean (95% con dence interval) or n (%). P value by chi-square test. Abbreviations: WC; waist circumference.