Dental amalgam is a mercury-based filling containing approximately 50% of metallic mercury (Hg0). Human placenta does not represent a real barrier to the transport of Hg0; hence, fetal exposure occurs as a result of maternal exposure to Hg, with possible subsequent neurodevelopmental disabilities in infants. This study represents a substudy of the international NIH-funded project “Early Childhood Development and polychlorinated biphenyls Exposure in Slovakia”. The main aim of this analysis was to assess the relationship between maternal dental amalgam fillings and exposure of the developing fetus to Hg. The study subjects were mother–child pairs (N=99). Questionnaires were administered after delivery, and chemical analyses of Hg were performed in the samples of maternal and cord blood using atomic absorption spectrometry with amalgamation technique. The median values of Hg concentrations were 0.63 μg/l (range 0.14–2.9 μg/l) and 0.80 μg/l (range 0.15–2.54 μg/l) for maternal and cord blood, respectively. None of the cord blood Hg concentrations reached the level considered to be hazardous for neurodevelopmental effects in children exposed to Hg in utero (EPA reference dose for Hg of 5.8 μg/l in cord blood). A strong positive correlation between maternal and cord blood Hg levels was found (ρ=0.79; P<0.001). Levels of Hg in the cord blood were significantly associated with the number of maternal amalgam fillings (ρ=0.46, P<0.001) and with the number of years since the last filling (ρ=−0.37, P<0.001); these associations remained significant after adjustment for maternal age and education. Dental amalgam fillings in girls and women of reproductive age should be used with caution, to avoid increased prenatal Hg exposure.
Mercury (Hg) is a toxic metal that naturally exists in several forms in the environment — metallic mercury (also known as elemental mercury), inorganic mercury (in the form of mercury salts) and organic mercury (methylmercury being the most common). The primary target organs for Hg toxicity represent the central nervous system and the kidney (WHO, 1991; Echeverria et al., 1998; Mason et al., 2001).
Dental amalgam is a mercury-based filling mixed with silver (35%), tin (9%), copper (6%) and zinc (trace amounts); it contains approximately 50% of metallic mercury (Hg0). The mercury is bound within the amalgam, but small amounts in the form of Hg0 are continuously released from the surface of the filling into mouth air and saliva due to corrosion, grinding motions or chewing. Average estimates of the amount of mercury released from dental amalgams range from 3 to 17 μg/day, depending on the total number of amalgam fillings (Lorscheider et al., 1995; Agency for Toxic Substances and Disease Registry, 1999). Mercury vapor from amalgam fillings leaks over time (Vimy et al., 1990; Takahashi et al., 2001; Pizzichini et al., 2003), with 80% of the inhaled Hg0 entering the blood via lungs. The biological half-life of mercury in the whole human body is approximately 40 days for inorganic Hg and 70–80 days for methylmercury (MeHg) (WHO, 2000).
Susceptibilities of children to the adverse health effects from Hg exposure may change over time, depending on the developmental stage. The first exposure of children to environmental pollutants takes place during prenatal development via transplacental transport (Bearer, 2000). An increase in cord blood Hg, placental Hg or breast milk Hg in relation to the number of maternal amalgam fillings was shown in several studies (Drexler and Schaller, 1998; Vahter et al., 2000; Ask et al., 2002; Bjornberg et al., 2003, 2005a; Ursinyova and Masanova, 2005). The human placenta does not represent a real barrier to the transport of Hg0 and MeHg, both known to be neurotoxic agents mainly in the early stages of development; hence, fetal exposure occurs as the result of maternal exposure to Hg, with possible subsequent neurodevelopmental deficits in infants (Ask et al., 2002; Debes et al., 2006).
Significant associations between the number of amalgam surfaces and the Hg concentrations in human organs were found in adults (Weiner and Nylander, 1993). Takahashi et al. (2003) demonstrated in an animal model that mercury vapor released from the amalgam fillings was distributed to maternal and fetal organs in a dose-dependent relationship to the number of amalgam fillings.
Potential toxicity from exposure to mercury from dental amalgam fillings is the subject of a long-standing debate by scientists and health officials worldwide (Clarkson, 2002; Counter and Buchanan, 2004); to date, the results of scientific studies remain inconclusive and conflicting. Some authors found a relationship between amalgam exposure and adverse health effects in adults, mainly in the nervous system (Echeverria et al., 1998). In contrast, other authors have emphasized that exposure to Hg from amalgam dental fillings is at low levels, and have argued that despite long-term exposure, it is insufficient to adversely affect human health (Saxe et al., 1995; Loftenius et al., 1998; Dodes, 2001; Factor-Litvak et al., 2003; Bates et al., 2004; Hujoel et al., 2005; Kingman et al., 2005; DeRouen et al., 2006). The majority of these studies focused on amalgam exposure in adult populations and not in newborns/infants.
This analysis represents a substudy of the international NIH-funded project “Early Childhood Development and polychlorinated biphenyl (PCB) Exposure in Slovakia” (Hertz-Picciotto et al., 2003), launched in 2001. The parent project is concerned primarily with the relationship between prenatal/postnatal exposures to PCBs, and the health and development of infants living in two districts of eastern Slovakia — Michalovce and Svidnik/Stropkov, about 70 km apart. Toxic metals mercury and lead are being considered as possible confounders/effect modifiers of the neurotoxic effects of PCBs.
The main objective of this study was to assess the relationship between maternal dental amalgam fillings and Hg exposure in the developing fetus/newborn. Cord blood Hg concentrations were used as the biomarker of prenatal exposure to Hg.
Material and methods
The study subjects were mother–child pairs (N=99) enrolled in the study “Early Childhood Development and PCB Exposure in Slovakia” at the time of delivery in the main district hospital from each region — Michalovce and Svidnik.
After the administration of informed consent, pregnant women were enrolled into the study if they met the following eligibility criteria: women at least 18 years old or older, free of serious illnesses during pregnancy, parity between 0 and 4, and having lived in the district for five years or more. Interviewer-administered questionnaires were used to collect demographic data, information on pregnancy and lifestyle (e.g. smoking, alcohol consumption), maternal diet, sociodemographic status, information on the number of amalgam dental fillings in mother, and her age when she received her first and her most recent dental fillings. These interviews were conducted by trained study staff during the mother's hospital stay within a few days of delivery.
The project was approved by the Ethical Committees of the Slovak Medical University, Bratislava, Slovakia, and the School of Medicine at the University of California, Davis, USA.
As a part of the parent study protocol, maternal blood specimens (1 ml) were collected by venipuncture into heparinized vacutainers (S-Monovette LH-Metall-Analytik tubes; fy SARSTEDT) and stored frozen (−18°C) until laboratory analysis. Cord blood specimens (1 ml) were collected after delivery of the placenta by needle aspiration of the fetal vessels on the fetal placental surface; dripping the placental blood directly into laboratory tubes was allowed if there was a free flow of blood. Cord blood samples were processed by the same method as the samples of maternal blood.
The determination of Hg concentrations was performed by atomic absorption spectrometry with amalgamation technique using the advanced mercury analyzer AMA-254 fy. Altec (Czech Republic). The principle of the method was published elsewhere (Ursinyova and Masanova, 2005).
Mercury concentrations were measured directly in heparinized blood samples (150 μl blood). The method of aqueous calibration curve was used with duplicate measurements performed on all samples. The analytical steps were set as follows: 170, 120 and 40 s for drying, decomposition and cuvette clearing, respectively. The 3-sigma detection limit was 0.10 μg/l in the whole blood. An internal quality control was performed in every 10 measurements with the reference material SeronormTM Trace Elements Whole Blood — Level 1 (Sero AS, Norway). Percent recoveries of mercury in the reference material were found to be in the range 91–107%. Validation of the method was also realized by successful participation in the SEKK proficiency-testing survey (Czech Republic) for trace elements in whole blood. All laboratory procedures were carried out under the QA/QC.
Statistical data analyses were performed using SAS 9.1 and STATA 6.0. Shapiro–Wilk W-test was used for the assessment of normality in data distributions. Since Hg concentrations were not normally distributed, besides arithmetic mean, we present geometric mean with 95% CI, median and range to summarize Hg concentrations in the maternal and cord blood samples. Spearman's correlations, contingency tables, Kruskall–Wallis and Wilcoxon rank-sum tests were used to assess the relationship between covariates in the study and both the primary predictors and the outcomes. The time since the placement of the most recent amalgam dental filling was calculated as the difference between the actual age of the mother at delivery and her age when she had received the latest dental filling. The highest maternal age in the study was 42 years; this value was assigned for time since the most recent dental filling to mothers who had no amalgam dental fillings. For the analysis of the effect of the number of maternal amalgam dental fillings on prenatal Hg exposure, dummy variables were created for three categories: 0–2 fillings, 3–7 fillings and more than 7 fillings. Maternal education was used as a dichotomized variable — high school without graduation or lower, versus high school graduation or higher.
Final multiple regression models were built using log-transformed cord blood Hg concentrations as the outcome. Separate models were fit to assess the number of maternal dental fillings and the role of time since placement of the most recent dental filling as predictors of interest. Reciprocal values of the time since placement of the latest amalgam filling (1/x) were used in the multiple regression analysis to model its effect on prenatal Hg exposure.
Ninety-nine mother/child pairs participated in this study, representing both study regions — Michalovce (N=52) and Svidnik/Stropkov (N=47). Characteristics of the study population are presented in Table 1. The number of amalgam fillings ranged from 0 to 20 (mean 5.6), with 10 women having had a new filling placed within the past 12 months. Maternal age ranged from 18 to 42 years (mean 25.7 years). Average number of years of education was 11.7 years, 55% of women had graduated from high school or had a higher education level. Out of 99 women, 76% were from Slovak or other Eastern European and 24% belonged to Romani ethnicity. Fifty-six percent of the women were primiparas, 47% of had women smoked before pregnancy and 22% of them continued to smoke during pregnancy. Fish consumption in this study population was very low, with only 29% of women indicating that they ate some fish during pregnancy (data not shown).
The medians, range, arithmetic and geometric means, and 95% CI of Hg levels (μg/l) in cord blood and maternal blood samples are shown in Table 2. The median values were 0.63 μg/l (range 0.14–2.9 μg/l) and 0.80 μg/l (range 0.15–2.54 μg/l) for Hg concentrations in maternal and cord blood, respectively. We found a significant positive correlation between maternal and cord blood Hg concentrations (Figure 1); higher Hg concentrations were found in cord blood, as compared with maternal blood. On average, cord blood Hg concentrations were 1.2 times higher than Hg concentrations in maternal blood (5th and 95th percentiles 0.56 and 2.2 times higher, respectively).
Bivariate analyses showed a positive association between the number of maternal fillings as a continuous variable and the levels of Hg in the cord blood (ρ=0.46, P<0.001) and an inverse relationship between the levels of Hg in the cord blood and time since the latest filling was inserted (ρ=−0.37, P<0.001). The strength of the association was similar for the number of dental fillings and concentrations of Hg in maternal blood (ρ=0.46, P<0.001). The median Hg levels in maternal blood samples were 0.41 and 0.67 μg/l, for mothers without amalgam fillings, and mothers having at least 1 amalgam filling, respectively (P=0.0028). No significant effects of district or selected environmental characteristics of maternal residence (e.g. distance to an industrial factory, or dump site) were found on Hg levels in maternal or cord blood samples.
Further variables chosen for the preliminary model of predicting cord blood Hg concentrations were maternal education, maternal age, being primipara, active/passive exposure to smoking during pregnancy and ethnicity. Selection of variables of interest was based on previous knowledge regarding the factors influencing blood Hg concentrations and bivariate associations with predictors and outcome observed in these data. However, the directed acyclic graph showed that adjustment for age, education and ethnicity was sufficient to control confounding (Figure 2). On the basis of the strong associations between ethnicity and maternal age, and between ethnicity and maternal education, the variable ethnicity was not used in the final model. The same approach was used for the assessment of the relationship between time since the latest dental filling insertion and cord blood Hg concentrations.
Results of the final multivariate linear regression models are shown in Tables 3 and 4. After adjustment for maternal age and maternal education, both the number of maternal amalgam dental fillings and the effect of the time since the most recent dental filling insertion remained significant predictors of cord blood Hg concentration (P=0.003 and P=0.0462, respectively). We did not find any effect of the number of maternal dental amalgam fillings on the blood Hg levels in 6-month-old children (data not shown).
The aim of our study was to evaluate the association between the number of maternal dental amalgam fillings and prenatal exposure to Hg, using cord blood Hg levels as the biomarker of prenatal Hg exposure.
The strongest predictor of cord blood Hg levels was the concentration of Hg in maternal blood, confirming previous reports (Vahter et al., 2000; Bjornberg, 2005). Levels of Hg were higher in cord than in maternal blood, also in accordance with other studies (Drasch et al., 1994; Bjerregaard and Hansen, 2000; Ramirez et al., 2000), indicating partial trapping of Hg in fetal tissues, once Hg was transferred from maternal side through the placenta to the fetal side.
The Hg concentrations in cord and maternal blood in our study were approximately 2–10 times lower than those observed by Bjerregaard and Hansen (2000), Vahter et al. (2000) or Walker et al. (2006). This is likely the result of differences in dietary habits and other characteristics of the populations studied; residents of Arctic Canada, Greenland and presumably Sweden too are more likely to eat seafood — especially predatory fish and marine mammals. Seafood in the diet is the most important source of MeHg. In our study, information on the speciation of Hg was not available; but in general, the diet of the Slovak population contains only low amounts of fish and seafood. In the women enrolled in this study, only 29% reported consuming fish during pregnancy and no association between fish intake in pregnancy and maternal, or cord blood Hg concentrations was found. Given that the Hg level in whole blood reflects current inorganic and organic mercury exposure (Loftenius et al., 1998) and that exposure to organic Hg from fish consumption was minimal in our population, Hg exposure from amalgam dental fillings likely represented a major contribution to the total Hg concentration. The relatively strong association of the number of fillings and cord total Hg supports this conclusion; furthermore, the strength of the association was similar for the concentrations of Hg in maternal blood, and significant difference was observed between the median Hg levels in maternal blood samples for mothers without amalgam fillings and mothers having at least 1 amalgam filling, with no significant effect of other environmental characteristics on Hg levels in maternal or cord blood samples.
Final multivariate analyses included adjustment for maternal education and age. Although the effects of other variables such as ethnicity, parity and active/passive exposure to smoking during pregnancy on prenatal Hg exposure were examined (Figure 2), they were excluded from the final model, since controlling for maternal age and education was shown to be sufficient to control confounding, and ethnicity was strongly correlated with both maternal education and age.
On the basis of the associations among the selected variables in the model, ethnicity, smoking and parity appear to be proxy markers for other unspecified socioeconomic factors rather than covariates with direct effect on the cord blood Hg levels, or the number of maternal amalgam dental fillings in our study. Variable ethnicity had two categories — Slovak or other Eastern European (e.g. Hungarian, Czech, Polish) and Romani and it was a determinant of all variables in our model. In general, Romani population is characterized by living in poorer hygiene conditions, having lower socioeconomic status, lower level of education and a higher prevalence of smoking (Koupilova et al., 2001). In our study, being Romani was strongly associated with having lower age and education, a smaller number of maternal amalgam fillings, lower concentrations of Hg in cord blood, and higher parity and smoking. Smoking exposure showed a negative association with both cord blood Hg levels and the number of maternal fillings in bivariate analyses; and at the same time, it was strongly associated with the level of maternal education, younger age and Romani ethnicity, suggesting the indirect effect of this variable on the predictor and outcome of interest; hence, controlling for the level of education and age helped control for smoking exposure at the same time (Figure 2). The same reasoning was used for the variable parity.
Taking into consideration the biological half-life of Hg and given that Hg exposure from dental fillings is highest around the time of placement of the filling (Pleva, 1994), the most recently inserted dental fillings are expected to have a more important impact on prenatal Hg exposure than, for example, maternal fillings placed a few years ago. The further passage of Hg would be expected to have little impact. For these reasons, instead of a simple linear term, we used an inverse exponential relationship in the assessment of the role of the time since the latest amalgam placement.
Our study had several limitations. The sample size of the study population was relatively small and only the total Hg concentrations were measured, with no speciation of Hg during the analytical process. Data on maternal dental fillings were obtained using only a questionnaire, with no objective information from a dentist; hence, no information was available about the number and area of amalgam surfaces. Information on fish consumption was also based on questionnaire. Thus, given a relatively low Hg in blood in our population and our use of self-reported information about dental amalgam fillings, the significant positive relationship between the number of maternal amalgam fillings and cord blood Hg concentrations and the significant negative association between the years since the most recent amalgam dental filling insertion and cord blood Hg concentrations are more striking.
On the other hand, vaccination rate in Slovakia represents almost 99%, so the differences in maternal and cord Hg concentrations due to different level of vaccination were negligible. Ethylmercury is an organic Hg compound that has been used in the form of thimerosal as preservative in some vaccines. While the use of mercury-containing preservatives has declined in recent years, some vaccines containing thimerosal are still available in Slovakia (e.g. Diptheria-Tetanus-Pertusis, DTP) and represent one of the possible sources of Hg exposure in the general population.
In conclusion, we found a relationship between maternal amalgam fillings and prenatal exposure to Hg. Further studies need to focus on the adverse health effects of Hg from dental fillings, mainly in sensitive populations — for example, pregnant women and infants and to establish evidence-based guidelines for the use of dental materials during the reproductive age and pregnancy. In our cohort, children are being followed up to school age, and regular examinations include an evaluation of neurobehavioral development; hence, further assessment of the potential negative effect of prenatal Hg exposure will be possible.
We express appreciation for the funding received from the US National Institutes of Health, National Cancer Institute, grant # R01-CA96525 and Fulbright grant # 12/06 FC.
About this article
Environmental Monitoring and Assessment (2015)