Predictors of mercury, lead, cadmium and antimony status in Norwegian never-pregnant women of fertile age

Background The toxic trace elements mercury (Hg), lead (Pb), cadmium (Cd) and antimony (Sb) are transferred over the placenta to the fetus and secreted into the breastmilk. All four elements bioaccumulate in the body and as maternal age at delivery is increasing in industrialized countries, the burden of toxic trace elements in never-pregnant women of fertile age is of concern. Methods Healthy, never-pregnant women aged 18 to 40 years (n = 158) were recruited between June 2012 and March 2015 in Bergen, Norway. Clinical data were collected and non-fasting venous blood samples were analyzed for whole blood Hg, Pb and Cd and serum Sb by ICP-MS and related to diet and life style factors. Results In a multiple linear regression model, increasing age was associated with higher levels of Hg and Sb, but diet and life style factors were more important predictors. Median whole blood Hg was increased by a factor of 70 in women who had fish for dinner ≥1/week, compared to women who rarely or never ate fish (p<0.001). Alcohol intake was the strongest predictor for whole blood Pb, while use of tobacco was the strongest predictor for whole blood Cd. Being a vegetarian was associated with lower levels of both Hg and Sb. Conclusions As toxic trace elements tend to bioaccumulate in the body, increasing maternal age at delivery may represent a threat to the next generation. In a group of healthy Norwegian never-pregnant women, age contributed to Hg and Sb levels, but diet and life style factors were stronger determinants of whole blood Hg, Pb, Cd and serum Sb levels. Continuous public actions are needed to reduce modifiable and preventable sources of potentially deleterious toxins to minimize the exposure in children and fertile women.


Blood sampling and analyses
Non-fasting blood samples were obtained by antecubital venipuncture and collected into vacutainer tubes with EDTA and without additives approved for trace metal analysis (Terumo1, US). Five-hundred μL of whole blood and serum was aliquoted and stored at -80˚C until analysis. Whole blood Hg, Pb and Cd and serum Sb were analyzed by ICP-MS on Perkin Elmer DRC-e (Perkin Elmer1, USA) in standard mode [31].
Plasma levels of cotinine were assayed using a LC-MS/MS method [32]. A plasma cotinine level !85 nmol/L is commonly used as a cut off to define regular smokers [33], however use of smokeless tobacco is associated with similar cotinine levels [34].

Statistical analysis
Results are presented as mean and standard deviation (SD), compared by Student's t-test or Anova, and median and interquartile range (IQR), compared by Mann-Whitney U test or Kruskal Wallis test. Chi-square test was used for categorical data. Spearman correlations and multiple linear regression models were used to explore relationships between data.
Graphical illustration of the relationship between whole blood Pb and alcohol units consumed per week was obtained by generalized additive models (GAM).
The SPSS statistical program (version 23) and the packages "mgcv" inR1, version 3.3 (The R Foundation for Statistical Computing) were used for the statistical analyses. Two-sided p-values < 0.05 were considered statistically significant.
Most women, 124/158 (78%), had an omnivore diet, whereas 34/158 (22%) had used a vegetarian diet for a median of 36 months (IQR 16, 132), range 2-240 months. Fifteen of the 34 vegetarians (44%) were vegans and did not eat any animal products, including eggs and milk. Demographic data according to an omnivore or a vegetarian diet are given in Table 1.
Omnivorous were younger, had a higher educational level and were less likely to use tobacco compared to vegetarians (Table 1). A total of four women (2.5%) reported regular daily smoking, two from each diet group, however, based on plasma cotinine levels !85 nmol/ L, 17/158 (11%) women were defined as regular tobacco users. Use of alcohol ranged from 0 to 12 units per week, with no difference between the groups (Table 1). Two thirds (105/158) of the women reported regular use of micronutrient supplements, with no difference between omnivores and vegetarians (p>0.07) ( Table 1).
Toxic trace element levels according to age, diet and life style factors Levels of Hg, Pb, Cd and Sb according to age, diet, use of alcohol and tobacco are given in Table 2. Age was positively correlated to Hg (r = 0.18, p = 0.02), Cd (r = 0.27, p = 0.001) and Sb (r = 0.23, p = 0.004) by Spearman correlation, but only Cd and Sb levels were higher in women !25 years, when compared by Mann-Whitney to the younger age-group (Table 2).
Age remained a significant predictor for Hg and Sb in a multiple linear regression model, which additionally included type of diet, fish intake, alcohol intake, use of tobacco, education, BMI and use of omega 3/cod oil supplements (Table 3).
Vegetarians had lower levels of Hg and Sb compared to omnivores (Table 2), and this was evident also in the multiple linear regression model (Table 3). No differences in levels of toxic trace elements were seen between vegans and vegetarians (p>0.22, data not shown).
Fish intake was the strongest predictor for Hg in the multiple linear regression model (Table 3). Median whole blood Hg was increased by a factor of 70 in women who had fish for dinner ! 1/week, compared to women who rarely or never ate fish (Table 4). Alcohol intake was correlated to Pb by Spearman correlation (r = 0.37, p<0.001), and median Pb was significantly higher in women who drank 2 or more units of alcohol per week ( Table 2). There was a dose-response relationship between intake of alcohol in units per week and Pb, as shown by GAM corrected for age (Fig 1). Alcohol was also the only significant predictor for Pb in the multiple linear regression model (Table 3).
Regular use of tobacco resulting in plasma cotinine levels !85 μmol/L was associated with higher Cd and lower Hg levels ( Table 2). In the multiple linear regression models, tobacco use was strongly related to Cd and less to Sb.
A longer education was a negative predictor for Cd, while no significant associations to any of the trace elements were seen for BMI or use of omega 3 fatty acids/ cod oil supplements by multiple linear regression models (Table 3).

Discussion
In this group of never-pregnant women of fertile age, age was associated with higher levels of Hg and Sb, but the strongest predictor for Hg status was fish intake, alcohol intake for Pb, use of tobacco for Cd status, while the strongest predictor for Sb was an omnivore diet.

Strength and limitations
This was an observational study with self-reported clinical data, known to have disadvantages, as highlighted by the discrepancy between reported smoking status and plasma cotinine levels. However, as we were able to measure plasma cotinine this problem was diminished. This study included women of fertile age, ranging from 18 to 40 years, however, the majority of our population was below 30 years, which disabled us to fully investigate the effect of a more advanced age on toxic trace element status.

An omnivore versus a vegetarian diet
Published data concerning toxic element status in vegetarians compared to omnivores are somewhat conflicting [28,35,36]. In a non-smoking population, vegetarians had higher levels of Cd compared to omnivores [28], and both vegan and vegetarian diets have been proposed to be a risk factor for increasing the level of several toxic trace elements in the body [35]. However, in a Swedish study, a change from a mixed to a lactovegetarian diet for 3 months was associated with lower hair concentrations of mercury, lead, and cadmium [36]. We did not find any difference in Pb and Cd according to diet. An omnivore diet was a predictor for Sb, and the negligible Hg levels we found in our vegetarian population compared to omnivores with a high fish intake, confirms the notion that Hg levels are primarily the result of intake of fish and seafood [27] and therefore reduced in vegetarians.

Mercury
The level of Hg is reported to increase with age [37], confirmed by our data on women aged 18 to 40 years. Compared to published studies (Table 5), the median Hg value (4.91 nmol/L) in our population was low. Hg levels vary in different populations according to published studies  (Table 5). A considerable higher Hg level, geometrical mean (GM) 13.1 nmol/L, was reported in Korean women above 20 years in 2008 [41], while a lower level, median 3.74 nmol/ L, was found in a Czech mixed population aged 18-58 years in 2009 [39] (Table 5). The high level of blood mercury in the Korean study has been explained by a high seafood consumption in this population [41]. Fish consumption is generally high in Norway [47], but the lower median Hg level in our population might be due to a high percentage of vegetarians (22%). Omnivores with a high fish intake had a median Hg level of 7.0 nmol/L, comparable to Danish women of fertile age (geometrical mean (GM) 7.93 nmol/L) [43] (Table 5). Tobaccousers had also lower levels of Hg and conversely a lower intake of fish, compared to non-users of tobacco. Smoking is associated with overall unhealthy behavior, including poor nutrition [48], while eating fish and seafood is considered to be beneficial [49]. According to current Nordic recommendations, Norwegians should have an intake of 300-450 g fish per week, of which at least 200 g should be fatty fish, with no discrimination according to age and sex [50]. Fish provides important nutrients, it is however important to consider the fact that fish is also the most important contributor of Hg, known to have serious negative health effects, even in low doses [3].

Lead
The median Pb value of 0.04 μmol/L was low in our population and comparable to published data from Nordic female populations [38,41,43,44] (Table 5). Pb levels in premenopausal, non-pregnant women are primarily due to exposures to contaminated food or water [51]. Pb are commonly found in brewed tea [52], reported to be a popular drink among women [53], however, as we did not have any data on tea consumption, we were unable to evaluate this factor. Pb is also commonly found in red wine [54], which corroborate with our finding of alcohol intake being the only significant predictor for Pb level. According to 2014 data from the Norwegian Institute of Public Health, the alcohol consume has increased by approximately 40% in Norway during the last 20 years, particularly among women, who preferably drink wine [55].

Cadmium
We found a median Cd value of 1.53 nmol/L, which is low compared to other countries ( Table 5). As smoking has been associated with higher levels of Cd [25], this finding might be related to a low prevalence of smokers in our population. The use of snuff is becoming increasingly popular in Norway [56], but studies on how use of smokeless tobacco affect toxic element status are difficult to find [57]. We found a 63% higher median Cd level in regular users of tobacco, confirmed by cotinine levels, compared to non-users. The median cadmium level in our non-users of tobacco, 1.51 nmol/L, was about twice as high as the value of Finnish nonsmokers aged 31 years, 0.62 nmol/L, while the median level in our current tobacco-users, 2.47 nmol/L, was substantially lower than in the Finnish smokers, 6.04 nmol/L [38].

Antimony
Our median Sb levels (30.1 nmol/L) was comparable, but higher than in Chinese females aged 23-64 years in a study from 2014,where they found a geometric mean (SD) of 14.78 (± 8.9) nmol/L [46] (Table 5). The most significant predictor for Sb in our population was an omnivore diet, followed by age and use of tobacco. Sb has been found in all food groups, with the highest concentrations in dairy products [26]. Sb levels are also reported to be higher in smokers [58], as confirmed by our results.
Occupational exposure in electronic waste recycling was the most important source of Sb in the Chinese population, while dietary exposure was considered to be low [58]. Data on environmental exposure sources of Sb is scarce and further studies are obviously needed. Sb is frequently found in mobile phones [58], a popular electronic article, however, whether the use of mobile phones may contribute to higher serum levels is unknown.
Sb has been reported to cause neurotoxicity, liver and respiratory disease and has been associated with an increased cardiovascular risk [1,10]. Sb may cross the placenta [1], and is also found in breastmilk [11]. It has been suggested that antimony trioxide may interfere with fetal development [1,59] and higher concentrations of Sb have been found in cord blood in pregnancies with an adverse outcome [18].

Implications
Element interactions may amplify the effect of even low dose concentrations of toxic elements [4] and safe limits are considered to be unlikely [60]. As toxic elements tend to bioaccumulate in the body [37,61], a higher maternal age may have an impact on the burden of environmental toxins transferred to the next generation. However, we find that fish intake, alcohol consumption and use of tobacco, are more important determinants for Hg, Pb and Cd status than age in females from 18 to 40 years.
As the only possible way to protect a fetus is to minimize the exposure of toxic elements in fertile women [62], continuous public actions are needed for reducing these modifiable and preventable sources of potentially deleterious toxins.

Conclusion
As toxic trace elements tend to bioaccumulate in the body, increasing maternal age at delivery may represent a threat to the next generation. In this study of Norwegian never-pregnant women aged 18 to 40 years, older age was associated with higher levels of Hg and Sb, but the main predictor for Hg levels was fish intake, for Pb alcohol intake, for Cd regular use of tobacco, while an omnivorous diet was the strongest predictor for Sb levels. As these are modifiable and preventable sources of harmful toxic elements in fertile women, public consideration and actions are needed to minimize the potential transfer of toxins to the next generation.