Thyroid hormones and neurobehavioral functions among adolescents chronically exposed to groundwater with geogenic arsenic in Bangladesh

Abstract

1. Introduction

Geogenic arsenic (As) occurs in elevated concentrations in the Holocene alluvial aquifers of rural Bangladesh resulting in chronic As exposure in millions of people who heavily rely on water in aquifers for both drinking and cooking. As concentration in drinking water in Bengal Basin often exceeds the World Health Organization (WHO) and US-EPA safety standard of 10 μg/L (Biswas et al., 2012; Halder et al., 2012; von Bromssen et al., 2007). In addition, several other toxic elements such as manganese (Mn), chromium (Cr), and lead (Pb) have been found in groundwater in Bengal Basin (Frisbie et al., 2009; Halder et al., 2013; Rahman et al., 2013). However, considering a large 150 million As-affected global population and more frequent detection of unsafe level of As in well water relative to other metals and metalloids As remains the most important public health challenge in different parts of the world (Shankar et al., 2014).

It is anticipated that more than 3 million wells in Bangladesh have As above this threshold value (Raessler, 2018). In addition, rice and other crops, fruits, juices and poor air quality can lead to higher arsenic exposure (Halder et al., 2012; Wilson et al., 2012). Arsenic (As) is associated with increased risk for several cancers in addition to non-carcinogenic health effects in the respiratory, cardiovascular and nervous systems in adults (Chen et al., 2011; Chen et al., 2007; Hughes et al., 2011; Parvez et al., 2010). Both experimental and epidemiological studies have reported associations of chronic water arsenic (WAs) exposure with several neurodevelopmental functions such as intelligence, learning, memory and motor function in early life (Rodriguez-Barranco et al., 2013; Tyler and Allan, 2014). However, the underlying mechanisms of As-induced neurotoxicity have yet to be elucidated.

Thyroid hormone (TH) is important in the development and maturation of the central nervous system (CNS) from the pre-natal period to the middle of childhood (Rovet, 2014; Zoeller and Rovet, 2004). A very small change in maternal TH – even within the population reference range – may be associated with up to a 10% decrease in child intelligence (i.e. loss of approximately 5-10 IQ points) (Bath et al., 2013; Haddow et al., 1999; Henrichs et al., 2010; Li et al., 2010). On the other hand, As may have a direct effect on the thyroid function through the inhibition of enzymes involved in TH synthesis and disruption of TH receptor-mediated gene regulation (Davey et al., 2007; Davey et al., 2008). Epidemiological studies including the NHANES study have reported positive association of As exposure with thyroid stimulating hormone (TSH) and negative associations of As with free thyroxine (fT₄) and total triiodothyronine (tT₃) (Meltzer et al., 2002; Molin et al., 2017); (Meeker et al., 2009);(Ciarrocca et al., 2012). Arsenic also triggers autoimmune thyroid disease resulting in elevated levels of biomarkers such as thyroid peroxidase antibody (TPOAb) (Brent, 2010; Langer et al., 2007). Thus, it is reasonable to hypothesize that As may exert a neurotoxic effect during child development by interfering with TH biomarkers such as fT₄, tT₃, TSH and TPOAb.

We conducted a pilot study in a group of As-exposed adolescent children for gathering preliminary data and laying the foundation for a future study with a larger sample size, which will eventually test whether TH mediates As-induced neurotoxicity. Mothers of these children are participants of an ongoing Health Effects of Arsenic Longitudinal Study (HEALS) in Araihazar, Bangladesh, which started in 2000 (Wasserman et al., 2007; Wasserman et al., 2004). These adolescent children (born after 2000) and their mothers have been consuming tubewell water since birth as evident in the literature (Wasserman et al., 2007; Wasserman et al., 2004). Longitudinal As exposure data for these adolescents was collected from HEALS database. Based on the data, it was possible to identify a subset of children who had either consistently high (i.e. above the WHO WAs standard of 10 μg/L) or low (i.e. below the WHO WAs standard) lifetime WAs exposure. We considered these adolescents as our potential study participants. During adolescence, several important components of the brain such as the frontal and limbic regions demonstrate stable performance in cognitive functions at adolescence (de Graaf-Peters and Hadders-Algra, 2006; Spear, 2013). Specific key areas of the brain that support working memory do not become functional until early adolescence as well (Sander et al., 2012). For these reasons, any measured neurodevelopmental outcome during this critical window is most likely to reflect the functional outcome during the remainder of the life course.

At the time of recruitment for the present study, all of our adolescent participants were between 15 to 17 years of age living in Araihazar, Bangladesh in the HEALS study area. We hypothesized that consistent WAs exposure from birth to childhood predicts the levels of four serum TH biomarkers through negative associations with total triiodothyronine (T₃) and free thyroxine (T₄), and positive associations with thyroid stimulating hormone (TSH), and antibodies to thyroperoxidase (TPOAb). Also, we tested the hypothesis that concurrent TH biomarker levels are associated with reduced neurobehavioral (NB) performance measured by a computer test battery, the Behavioral Assessment and Research System (BARS) (Rohlman et al., 2001; Rohlman et al., 2007a; Rohlman et al., 2003). Collectively, our long-term goal was to generate preliminary evidence regarding possible mediation effects of TH on the association between geogenic As exposure and NB outcomes, which may eventually lay the foundation for a larger study.

2. Materials and Methods

2.1. Study area and participants

The population in Araihazar, Bangladesh is exposed to a wide range of WAs with the majority of the HEALS participants consuming low-to-moderate doses (0-150 μg/L) of WAs (Figure 1). In the first three years of the study, the HEALS recruited 11,746 adults between the age of 18 and 75 years including 726 women who had children born between 2000 and 2002. At the time of our study conducted in 2017, these HEALS children were between 15 to 17 years of age. These children were enrolled in a prospective child development study when recruitment for our pilot study began. Past studies among children in the HEALS demonstrated strong correlations (r=0.75) between mother’s urinary arsenic (UAs) and children’s UAs, which were adjusted for urinary creatinine providing evidence that both mothers and children largely consume water from the same wells (Wasserman et al., 2007; Wasserman et al., 2004). This is because every mother brings water for all family members from a specific well, the only source of drinking water for the household. At the time of recruitment, WAs and UAs data were available for these mothers covering at least five critical stages of childhood: birth, age 3, 6, 9, and 12, which helped us identify the children who were consistently exposed to WAs either above or below 10 μg/L throughout childhood. We found 304 healthy adolescent children with consistent exposure to low WAs, i.e. WAs<10 μg/L and 252 children with consistent exposure to high WAs (WAs ≥10 μg/L) since birth. At the time of recruitment, all participants of this longitudinal child development study were between 15 to 17 years of age. We randomly selected 16 healthy adolescents from the high WAs group and equal number of participants from the low WAs exposure group to conduct this pilot study using a cross-sectional design. All initially selected participants went through physical examinations by a HEALS study physician and none of them was identified with any symptom of subclinical hypothyroidism. Arsenic guideline value of 10 μg/L of WAs recommended by the World Health Organization (WHO) and US Environmental Protection Agency (EPA) was chosen instead of the Bangladesh standard (i.e. 50 μg/L) to make the findings of the study relevant to other As-affected populations across the globe. To accommodate necessary field and laboratory costs for this pilot project, a small number of sample (n=32) was chosen. Children with disability, known iodine deficiency, diagnosed thyroid disease or other chronic diseases were excluded. The recruitment of adolescent participants took place between May to July of 2017. This study was approved by the Columbia University Medical Center Institutional Review Board (IRB), Indiana University IRB, and the Bangladesh Medical Research Council IRB. Parental informed consent and child assent were obtained once the mother agreed to allow her child to participate in the study.

Open in a separate window

Figure 1.

Concentrations of As in tubewells of HEALS participants in Araihazar, Bangladesh

Note: This should be a single-column fitting image

3. Results

4. Discussion

It is difficult to mitigate As contamination in drinking water in Bengal delta sediments, because arsenic gets dissolved into the aquifer via a biogeochemical process due to high amount of organics in the sediments (Raessler, 2018). Therefore, potential associations between consistent groundwater As exposure in early life, biomarkers of thyroid function and brain development hinted by our pilot study data have high public health significance. Arsenic neurotoxicity was evident in our study as the low As exposure group performed better when the key outcome variables measured by the BARS test battery were compared (Table 1). Although differences in NB performance between the two exposure groups were not always significant in our study, perhaps due to the small sample size, several previous larger epidemiological studies conducted in the same study area have already documented similar neurocognitive effects of WAs in children and adolescents (Parvez et al., 2011; Wasserman et al., 2004; Wasserman et al., 2018; Wasserman et al., 2011). We wanted to understand if TH modulation could mediate the relationship between As and NB performance. Overall, we observed that the associations between WAs exposure and TH biomarkers were non-significant after accounting for covariates. However, the directions and trend of associations were consistent with our hypothesis. In an unadjusted model, we observed significant linear relationship between As exposure and TPOAb (p=0.04) when age, sex and BMI were not taken into account. On the other hand, significant positive associations between TPOAb and latencies (i.e. response time) in simple reaction time and symbol digit tests and significant negative association between TSH and match to sample count are suggestive of adverse effects of elevated TSH and TPOAb on specific neurobehavioral functions.

Experimental studies suggest that As has direct effect on the thyroid function through inhibition of enzymes involved in TH synthesis and inhibition of signaling at the level of receptor-mediated gene regulation. This latter effect is similar to the effect on the other nuclear hormone receptor superfamily members (Davey et al., 2007; Davey et al., 2008; Meeker et al., 2009; Palazzolo and Jansen, 2008). Evidences from several epidemiological studies are consistent with this mechanism. For instance, an Italian study found that UAs in As exposed urban policemen was associated with increased TSH and decreased total T₃ and free T₄ when compared with rural outdoor workers (Ciarrocca et al., 2012). A pilot study on women who consumed fish contaminated with As also found negative correlations between blood As and plasma total T₃ and free T₄ (Meltzer et al., 2002) whereas in a cross-sectional study of US males there was a positive association between blood As and serum TSH after accounting for age, BMI and other covariates (Meeker et al., 2009). In an adult population in India (mean WAs exposure >202.7 ppb), significantly elevated thyroperoxidase autoantibodies were found in WAs-exposed group compared to the unexposed (Das et al., 2012). A study on rural Hispanic adults exposed to low WAs (i.e. 2-22 ppb) identified cumulative WAs as a significant predictor of hypothyroidism (based on TH levels and examination by physicians) after accounting for potential confounders (Gong et al., 2015). All these studies were conducted on adult participants. Our study is the first that examined the association between As exposure and TH in blood in an adolescent population. The specific age group has made our study unique as more TH are produced for rapid growth and sexual development during this life stage. The change in biological systems in adolescents makes them more susceptible to endocrine disrupting chemicals than the adults (McMullen et al., 2017).

There are several compelling reasons for considering adolescence age group for our study. Adolescence is an important period of brain development and maturation as both frontal and limbic regions of the brain demonstrate stable performance in cognitive functions at this development phase (de Graaf-Peters and Hadders-Algra, 2006; Spear, 2013). Although several subtle neurodevelopmental outcomes start to disappear after puberty (de Graaf-Peters and Hadders-Algra, 2006), some other functional deficits appear more significantly during brain maturation at middle or late adolescence (Hadders-Algra, 2002). Due to these reasons, neurobehavioral deficits measured during this critical window may be indicative of those during adulthood and beyond. In our study, we did not observe significant associations of total T₃, free T₄, and TSH with WAs, however, the directions of associations in the adjusted models are similar to the results of previous studies. Given that individual variations in serum TH, especially T₄ and T₃ in normal healthy subjects are relatively narrow in relation to the population variance (Andersen et al., 2002; Koulouri et al., 2013), TH levels in our sample might represent TH status at young age in the arsenic-exposed population in Bangladesh. Since all participants of the study had consistent WAs exposure, the concurrent WAs exposure also represented As exposure throughout all critical stages of child development.

Significant positive correlations between As and TPOAb in our study indicate a possible mechanism of As-induced thyroid disease, which was previously demonstrated by other environmental toxicants such as metals and polyhalogenated biphenyls (Brent, 2010; Kahn et al., 2014; Langer et al., 2007). Our data are also consistent with an adult study that demonstrated relationships between As exposure and autoimmune thyroid disease (Gong et al., 2015). Our study is the first to examine the possible effect of As on thyroid disease in adolescents by using the biomarker TPOAb. Although our study did not demonstrate significant associations between As and TPOAb after accounting for demographic covariates, perhaps due to a small sample size, the directions and magnitude of the associations still hinted at a potential effect of As in developing thyroid disease.

Sex is believed to influence TH concentrations although the precise age and sex dependent responses after exposure to hormone disruptors in the environment remain unclear. Several studies reported differential responses by the male and female participants (Clark et al., 2012; Roelfsema et al., 2009; Suzuki et al., 2012). Our data showed that adolescent boys had significantly higher total T₃ (b=0.26; p=0.03) and non-significantly higher free T₄ than the girls when As exposure in included in the models. Boys also exhibited lower TSH and TPOAb than the girls in adjusted models. Such relationships may be consistent with earlier studies that showed women to be more sensitive in TH action by expressing higher levels of TSH, and lower levels of T₃ and T₄ (Surks and Hollowell, 2007).

The findings of our study are not because of a greater proportion of adolescent participants with subclinical or overt hypothyroidism particularly in the high WAs exposure group. Only one out of 32 children had TSH above 10 μIU/ml and 27 children were within the most commonly used reference intervals of 0.4-4.0 μIU/ml (Chaker et al., 2017). Additionally, free T₄ ranged between 0.7-1.4 ng/dL is not indicative of a sample with hypothyroidism.

A body of literature suggests that As is a potent neurotoxicant. Arsenic easily crosses the placenta and is transferred to the fetus (Concha et al., 1998) and therefore As exposure appears to have strong neurocognitive effects (Tolins et al., 2014; Tyler and Allan, 2014). Among As-exposed pregnant women in Bangladesh, we have also found a significant association between maternal and cord blood As (Hall et al., 2007). We have already demonstrated adverse dose-response relationships between WAs and multiple measures of neurocognitive function including child intelligence, working memory and several other neuropsychological outcomes in the U.S. and Bangladeshi children (Parvez et al., 2011; Wasserman et al., 2014; Wasserman et al., 2007; Wasserman et al., 2004; Wasserman et al., 2011; Wasserman et al., 2016). These findings are consistent with our early studies in Bangladesh on both younger (i.e. 6-year old) and older children (10-year old) (Wasserman et al., 2007; Wasserman et al., 2004). Two of our more recent studies conducted in the same population found additional evidence in other neurobehavioral (NB) domains such as motor function and working memory in pre-adolescent children (Parvez et al., 2011; Wasserman et al., 2011). Despite consistent findings on associations between As and NB outcomes, the pathway via which As affects brain development remains unknown. Our objective was to explore if TH alteration by WAs is one of the possible pathways of As-induced neurotoxicity. For this purpose, we examined whether concurrent TH biomarkers in adolescents who have been consistently exposed to either high or low levels of WAs since birth are associated with neurocognitive functions at present time point.

The requirements for TH in the development and maturation of the central nervous system (CNS) extend from the pre-natal period to the middle of childhood (Rovet, 2014; Zoeller and Rovet, 2004). A very small change in maternal TH within reference ranges may be associated with up to a 10% decrease in IQ (i.e. approximately 5-10 IQ points) and other cognitive outcomes (Bath et al., 2013; Haddow et al., 1999; Henrichs et al., 2010; Li et al., 2010). TH is known to affect the development of the cerebral cortex, cerebellum, pons, striatum, thalamus, and basal ganglia (Gilbert et al., 2012), which control cognition, memory, learning, language development and motor function (Sanchez-Pena et al., 2010; Wang et al., 2009). Data from our study provide initial evidence that TH in adolescents may predict their NB performance at present time, however larger studies are warranted to establish the relationship.

In addition to some environmental chemicals, several nutritional factors and chronic disease conditions in liver, kidney and cardiovascular system also limit thyroid function (Asvold et al., 2007; Hepner and Chopra, 1979; Pearce, 2012; Sarne, 2000; Schussler et al., 1978; Sun et al., 2015; Tarim, 2011; Ye et al., 2014). Deficiencies of micronutrients such as iodine, iron and selenium can have negative impacts on the thyroid system in children (Brix et al., 2011; Triggiani et al., 2009). Urinary iodine below 20 μg/L in pregnant women is correlated with lower or undetectable serum free T₄, total T₃ and higher TSH (Zimmermann, 2009; Zimmermann et al., 2008). Although we did not collect data on these covariates, future large-scale studies should include these potential effect modifiers and confounding variables to assess the As-TH relationship more accurately.

5. Potential Limitations

This was an exploratory study with a small sample size limiting the statistical power for examining exposure-outcome relationships and testing the mediation hypothesis. However, our objective was to obtain preliminary data providing hints of associations between As exposure, TH and neurodevelopmental outcomes. Our preliminary findings would lay the foundation for a larger study, which would eventually have adequate statistical power to test the potential mediation effects of TH biomarkers on As-induced neurotoxicity. Given that our sample size was small, it was therefore important to assess if the directions of association of As with TH biomarkers (positive direction with TSH and TPOAb, and negative direction with T₃ and T₄) were visible in our sample. Since directions of relationship in our preliminary data were consistent with the hypothesis our findings create the opportunity to eventually test the hypothesis regarding the TH-dependent neurotoxicity of As using a large sample size in future.

There are several methodological limitations in our study. Drinking water is the primary source of As exposure in this population and therefore, As in household well-water was used as an exposure variable in our analysis. Although seafood is not a major source of As in rural Bangladesh other dietary sources of arsenic such as rice and vegetables could contribute to total As exposure in this population (Kippler et al., 2016). The relative contribution of dietary As to overall As exposure can be more significant when the WAs concentration is below the Bangladesh As standard of 50 μg/L (Kile et al., 2007). To overcome this limitation, we have used UAs, a measure of internal exposure biomarker that takes contributions of both water and dietary As into account for the adolescent participants. Recent studies conducted in Bangladesh demonstrated that WAs remained as the primary and most important source of As in most rural households (Khan et al., 2015; Kile et al., 2007), which is also evident in our study sample due to a high and significant correlation between WAs and UAs. We have previously analyzed arsenic metabolites in HEALS participants in Araihazar, and did not find arsenobetaine, a dietary source of As, a meaningful component of total UAs (Ahsan et al., 2007; Chen et al., 2011).The use of UAs, a biomarker of arsenic exposure also helped us minimize the limitation associated with undocumented information regarding multiple sources of water As. It was highly likely that participants of this study might have been consuming As-containing water from school wells or other neighborhood wells.

Absence of a third category of As exposure between the WHO and Bangladesh standard (i.e. 10-50 μg/L of WAs) was another limitation. It would have been interesting to know if this specific range of moderate exposure had any influence on TH and neurobehavioral outcomes when compared with <10 μg/L of WAs. However, such category was not statistically meaningful as there were only 2 participants in this exposure range.

Lack of As speciation data is another weakness of the study. Both trivalent (As^III) and pentavalent (As^V) forms of iAs are present in drinking water in Bangladesh although As^III was identified as the dominant form in groundwater and rice samples (Halder et al., 2014; Zheng et al., 2004). In vertebrates, As^III and As^V undergo multiple oxidation and methylation reactions producing monomethylarsonic acid (MMA) and then dimethylarsinic acid (DMA), which are excreted via urine. MMA is considered more toxic and high MMA% and iAs% in biological samples were reported to be associated with increased risk of skin lesions, skin and bladder cancers and cardiovascular outcomes in adults (Ahsan et al., 2007; Gao et al., 2011; Huang et al., 2008; Kile et al., 2011; Pierce et al., 2013; Valenzuela et al., 2005). A small number of studies reported associations of elevated MMA% and reduced DMA% with developmental delays and poor health of school children in a dose-dependent manner(Chen et al., 2013; Hsieh et al., 2014; Hsueh et al., 2016). However, our two studies in Bangladeshi children did not find an effect of MMA% on intelligence even though associations were reported for total BAs and UAs (Wasserman et al., 2004; Wasserman et al., 2018). In future, we plan to investigate these possible effects of methylated As species on TH and neurodevelopmental outcomes.

The potential influence of other neurotoxic contaminants such as manganese (Mn) in drinking water in adolescent participants cannot be ruled out. We did not measure Mn in water or biological samples for the present study, which may be considered as a potential limitation. At the same time, it needs to be noted that our previous studies on younger HEALS children did not observe synergistic effect of simultaneous As and Mn exposures (i.e. Mn by As interactions) on multiple neurobehavioral outcomes (Wasserman et al., 2007; Wasserman et al., 2011). Therefore, any potential influence of water Mn on our findings was very unlikely.

Being a cross sectional study it hinders cause and effect inferences although we argue that the direction of the relation is exposure to outcome. It is very unlikely that the levels of TH biomarkers can influence arsenic exposure or NB outcomes can influence TH. Our future plan is to re-assess the associations using a larger sample size that would be sufficient enough to examine if TH can mediate the associations between WAs exposure and NB outcomes. Furthermore, we will continue to follow that larger cohort to examine NB performance longitudinally. Our study was conducted in families primarily with marginal economic condition living in rural Bangladesh and therefore results may be generalizable only to communities with similar rural sociodemographic characteristics.

Although the number of subjects included in this study was limited, it did allow us to perform analyses for blood levels of TH biomarkers on a single 96 well plate. As a result, there was no systematic bias in the relative levels of these analytes. That is, the absolute values are less important than the difference among subjects and this was not biased by plate differences in the assay.

Another potential limitation is the absence of data regarding iodine status among the participants. Iodine deficiency has a known impact on the thyroid system in children (Brix et al., 2011; Triggiani et al., 2009). In turn, iodine insufficiency is related to a number of neuropsychological outcomes in young age (Chen and Hetzel, 2010; Gilbert et al., 2012; Qian et al., 2005). Future studies should assess iodine status of the adolescents to examine the effect of this nutritional factor on the associations between As, TH and NB outcomes.

In vitro and in vivo observations indicate that As may interfere with TH action in a tissue-specific manner (Davey et al., 2007; Davey et al., 2008; Meeker et al., 2009; Palazzolo and Jansen, 2008). Because of the effect of As on TH metabolism and receptor activation, it is possible that As interferes directly with mechanisms of TH actions in target tissues even in the absence of change in these hormone biomarkers under a variety of physiologic conditions (Salvatore et al., 2014; Watanabe et al., 2006). Therefore, it is possible that serum TH may not necessarily predict TH action in tissues. Although our findings are suggestive of As-induced TH change in blood, tissue-specific disruption of TH by WAs is also possible, which may in turn lead to decline of NB functions. This alternative mechanism, however, was beyond the scope of this project.

6. Conclusion

Overall, data from our study hinted possible TH disruption in childhood by geogenic As. Significant correlations of both WAs and UAs with TPOAb, non-significant positive relationships of As and with TPOAb and TSH, and negative relationship with free T₄ in the adjusted statistical models also emphasized the need of larger epidemiological studies for conclusive evidence about these associations. Furthermore, significant associations of TSH and TPOAb with a number of NB outcomes are consistent with our hypothesis. From risk prevention point of view, this is very important, because thyroid hormones play vital roles in determining the functions of the brain and therefore, insults in early life may have adverse effects on brain development in adulthood. Our future plan is to conduct epidemiological studies with larger samples to examine As-induced neurotoxicity mediated by TH disruption in early life. This line of research may provide a model that can be extended to additional risk factors for the brain, as well as for many other disorders aggravated by thyroid disrupting chemicals. If As-induced TH disruption is elucidated as a possible mechanism of neurotoxicity, it will help develop policies or therapeutic options to minimize early life adverse effects in the developing central nervous system.

Acknowledgments

Footnotes

References