The analyses presented here from CHARGE Study participants focus on blood Hg concentrations in children who have already been diagnosed with an ASD. They illuminate several facets of the blood Hg × AU/ASD relationships. First, fish consumption, a major contributor to human body burdens of methylmercury, was the primary predictor of total blood Hg in this population. Second, fish consumption was less prevalent in cases compared with controls. Consequently, when we ignored variation in fish intake (hence assuming it had no association with diagnosis), blood Hg in children with AU/ASD was significantly reduced compared with that of TD controls.
The third observation, however, is that after adjusting for dietary, medical, pharmaceutical, and dental sources of Hg, we found no difference in blood Hg comparing children with AU/ASD and TD controls, indicating that for a given exogenous exposure, cases did not differ from controls. Additionally, using specialized statistical methods to parse out the influence of behaviors that are affected by the autism diagnosis on blood Hg levels, we determined that, above and beyond the indirect impact on blood Hg mediated via fish intake, any direct effect of AU/ASD on blood Hg was small (~ 12% reduction). Finally, analyses restricted to non-fish-eating children demonstrated that cases had similar concentrations of blood Hg as did controls, after adjusting for non-fish Hg sources. Thus, three distinct analyses confirm that children with AU/ASD have blood Hg levels comparable with those of age-matched TD controls.
Notably, because half-lives of methylmercury in blood and whole-body inorganic Hg range from 60 to 90 days (Clarkson et al. 2003
), these measurements cannot address whether Hg exposures in either the prenatal or early postnatal period play an etiologic role in autism. An alternate related hypothesis posits that children with autism have abnormal metabolism or inadequate excretion of Hg and that they sequester Hg in the brain, resulting in greater susceptibility to its developmental neurotoxicity. Because the present study used a single time measurement of Hg in blood, we could not directly evaluate excretion or determine tissue levels of Hg in the central nervous system. An autopsy study of general population adults showed strong correlations between methylmercury in blood and in the occipital cortex and between the number of dental amalgams and concentration of inorganic Hg in brain (Bjorkman et al. 2007
). In monkeys, brain:blood ratios of total Hg were about 2.5 for those exposed to methylmercury (Burbacher et al. 2005
). Because only 5% of body burdens of Hg are estimated to be in circulation (Burbacher et al. 2005
; Stinson et al. 1989
), reliable conclusions about distribution are not possible from one-time observational measurements in blood.
A further limitation was our lack of data on dietary factors other than fish, such as selenium or long-chain polyunsaturated fatty acids, which may mitigate effects of Hg (Beyrouty and Chan 2006
; Ralston et al. 2008
; Strain et al. 2008
). Additionally, exposures to environmental Hg from ambient air pollution or other sources were not controlled in this analysis. Thus, our findings, that TD children with dental amalgams who chewed gum or ground their teeth had significantly higher blood Hg levels whereas AU/ASD children in that category did not, could be a result of unmeasured confounders that disproportionately affected AU/ASD children, or simply of the small number with this combined exposure (8 ASD, 4 DD, and 7 TD).
Strengths of this study include a relatively large sample size, recruitment of population-based cases and controls, and confirmation by trained clinicians of all diagnoses or lack thereof (in controls). The thorough collection of individual-level data with the assessment of multiple dietary, home, and medical sources of Hg is unique in the field of autism. Maternal reports of recent fish consumption were highly predictive of the child’s blood Hg level, suggesting strong validity of these questionnaire items. Further, we carried out a systematic approach to adjustment for selection bias, controlled confounding from numerous sociodemographic and other factors, and analyzed variability across developmental groups in blood Hg independently of fish intake. For these reasons, our results are likely to be both valid and generalizable.
In further support of generalizability, blood Hg values in these 2- to 5-year-old children with AU/ASD were similar to those reported for the U.S. NHANES nationally representative sample of 1- to 5-year-olds in 1999–2002 (Centers for Disease Control and Prevention 2004
). In the NHANES sample, the geometric mean was 0.33 μg/L, the median 0.26 μg/L, and the interquartile range 0.10–0.61 μg/L (Centers for Disease Control and Prevention 2004
); corresponding values in this study were 0.28, 0.23, and 0.10–0.57 μg/L (for this comparison, we used the NHANES detection limit of 0.14 to compute geometric means).
We presented several analyses because the underlying causal model is not certain. Current fish consumption and other Hg exposures might represent surrogates for earlier exposures or demographic factors that influence risk for AU/ASD or likelihood of diagnosis (), might be influenced by the child’s developmental status (), or both. Because our goal was to address whether children with a particular diagnosis exhibit differences in circulating Hg, adjustment for intake was essential (model 1, ). This model confirmed no differences in blood Hg in AU/ASD versus TD controls.
That children with DD had lower concentrations of blood Hg, whether or not we adjusted for fish consumption or restricted analysis to non-fish eaters, may indicate either a chance finding in the smaller sample of DD children or selection bias beyond that associated with socioeconomic factors. Nevertheless, because metabolic disorders are known to accompany numerous developmental conditions, further research on Hg kinetics, including genetic and environmental influences, may be warranted.
Few studies on autism have considered dental amalgams, personal care products, or diet as sources of Hg. Two reports related autism risk to ambient Hg pollution as a potential etiologic factor (Palmer et al. 2006
; Windham et al. 2006
) but did not have individual-level dietary data. Similar to our finding, investigations of island populations (Grandjean et al. 1992
) and the U.S. NHANES data (Mahaffey et al. 2004
) have reported fish and seafood consumption to be a major contributor to Hg measured in biospecimens. In a longitudinal study of 67 persons with initially elevated blood or hair Hg, reduction in fish consumption was followed by a steep decline in blood Hg level (Hightower and Moore 2003
). In nursing mothers from Germany, breast milk concentrations of Hg were strongly influenced by fish consumption, and maternal amalgams were a major source of infant Hg shortly after birth (Drexler and Schaller 1998
). A study in eastern Slovakia, where fish consumption is very low, demonstrated higher cord blood Hg concentrations with a greater number of maternal Hg amalgams, or a shorter time since the most recent filling; by 6 months of age, maternal amalgams were no longer determinants of the infant’s total blood Hg (Palkovicova et al. 2008
). Exposure from Hg amalgams occurs through inhalation of vapor released into the oral cavity, and both organic and inorganic Hg concentrations in saliva (Leistevuo et al. 2001
) and urinary Hg (Hansen et al. 2004
) are elevated in direct relation to the number of amalgam tooth surfaces.
The strong associations of maternally reported fish consumption and dental amalgams with blood Hg suggest that these questionnaire items provided valid information on recent exposures. Three types of fish each contributed independently to blood Hg. Interestingly, although few children had Hg amalgams, those who did and who also either chewed gum or had bruxism appeared to have experienced sufficient release of inorganic Hg to be measurable in blood. This finding is consistent with a report on adult chewers of nicotine gum (Sallsten et al. 1996
) and an investigation of amalgam Hg releases into chewing gum (Hansen et al. 2004
). Use of two over-the-counter thimerosal-containing products—nasal sprays and earwax removal formulations—was initially suggestively associated with elevated blood Hg but not in fully adjusted models. Misclassification of this exposure (relative to the timing of the blood draw) was possible because these products tend to be used intermittently and because some may not contain thimerosal.
The higher blood Hg in children whose mothers were born in neither the United States nor Mexico might have arisen from greater seafood consumption. These mothers came from all over the world. In the early phase of the study, we did not obtain the amount and timing of consumption of seafood, which contains per weight Hg levels about one-fourth those in fish (Mahaffey et al. 2004