Low level arsenic exposure, B-vitamins, and achievement among Uruguayan school children

Abstract

Introduction

Arsenic exposure is a global public health concern (George et al., 2014). Millions of people worldwide, including the U.S., are exposed to low levels of arsenic from water and food (Focazio et al., 2000; Naujokas et al., 2013). Food is a particularly important source of arsenic exposure when water arsenic levels are low, as it is in much of the U.S. (Davis et al., 2012; Rey deCastro et al., 2014; US Food and Drug Administration, 2014). We know that children exposed to high levels of arsenic can experience cognitive deficits (Rosado et al., 2007; von Ehrenstein et al., 2007; Wasserman et al., 2007; Wasserman et al., 2004; Wasserman et al., 2011). On the other hand, the effects of low-level arsenic on children’s neurobehavioral outcomes are poorly understood (Wasserman et al., 2014). We also have little evidence to inform how low-level arsenic exposure affects learning and school achievement.

Previous work on lead highlights the importance of studying effects of toxicants on school achievement (Amato et al., 2012; Miranda et al., 2007; Zahran et al., 2009; Zhang et al., 2013). As an outcome, academic achievement has direct use in public health decision making. Greater academic success facilitates student transition into independent adult lives. The relationship between arsenic exposure and school achievement has only been studied in children with fairly high level of exposure. One study in Bangladesh (water arsenic mean, range: 119.5, 0.1–1263.2 μg/L) showed no effect of arsenic exposure on children’s performance on a national scholastic test (Khan et al., 2012). Urinary arsenic (mean ± SD, 58.1 ± 33.2 μg/L) was also not associated with performance on a math achievement test in Mexican first-graders (Rosado et al., 2007). For low-level arsenic exposure (urinary arsenic, median, range: 11.9, 1.4–93.9 μg/L) , we observed no association with general intellectual abilities in first-graders (~7 year olds) in the Salud Ambiental Montevideo study in Uruguay (Desai et al., 2018). While academic achievement and intelligence are strongly correlated (Bartels et al., 2002; Jensen, 1998; Neisser et al., 1996; Sternberg et al., 2001), achievement depends on the child’s ability to utilize several cognitive domains, regulate behavior, and deploy social skills in peer and teacher interactions (Rimm-Kaufman and Chiu, 2007; Sánchez-Pérez et al., 2018). These abilities are not reflected in intelligence scores, hence the need to further evaluate the effect of low-level arsenic on academic achievement.

The role of arsenic methylation capacity, B-vitamin intake and status as potential modifiers of arsenic neurotoxicity remains largely unexplored (Desai et al., 2018; Hamadani et al., 2011; Hsieh et al., 2014). Once in the body, arsenic undergoes two methylation cycles that convert inorganic arsenic to monomethylarsonous acid (MMA), followed by dimethylarsinic acid (DMA) (Challenger, 1945). The proportion of MMA excreted in the urine reflects the body’s capacity to convert MMA to DMA. Higher %MMA and lower %DMA indicate lower methylation efficiency, which is a susceptibility factor for several arsenic-related health outcomes (Del Razo et al., 1997; Pu et al., 2007; Steinmaus et al., 2006; Vahter, 2001; Wu et al., 2006). Evidence from both observational studies and randomized clinical trials indicates that nutritional methyl donors, such as folate, influence the sequential methylation of inorganic arsenic to MMA and DMA, due to their role in synthesizing the universal methyl donor S-adenosylmethionine (Bozack et al., 2018; Gamble et al., 2006; Howe et al., 2014). Both intake and status of B-vitamins are associated with efficient arsenic methylation (Gamble et al., 2006; Gamble et al., 2005; Gamble et al., 2007; Hall et al., 2009a; Hall et al., 2009b; Howe et al., 2017; Steinmaus et al., 2005). Nevertheless, the role of B-vitamins in modifying arsenic-induced health outcomes is not well understood, particularly in populations with mandatory fortification.

The aim of the current study was to investigate the association between low-level arsenic exposure measured as the sum of urinary inorganic arsenic, MMA, and DMA, and academic achievement, thus expanding evidence beyond general intellectual functioning. We also tested arsenic methylation capacity (urinary %MMA) and dietary B-vitamin intake as potential effect modifiers. In a sub-sample of children, we explored whether serum folate and serum vitamin B-12 concentrations moderated the association between arsenic exposure and academic achievement.

The study site consisted of private elementary schools in Montevideo, Uruguay. Previous studies have shown that children in Montevideo are exposed to several toxic metals including lead, arsenic, cadmium, and manganese (Kordas et al., 2010; Mañay et al., 2008; Queirolo et al., 2010). Studies of young children have shown that exposure to metals co-occurs with anemia (Bradman et al., 2001; Wright et al., 1999). In Uruguay, commercially produced wheat flour is fortified with elemental iron and folate, but not other B-vitamins, since 2006. Assessing the role of low-level arsenic exposure on child development in such populations can provide insights into the potential role of diet in mitigating the adverse effects of arsenic.

Materials and Methods

Results

The median (range) U-As concentration was 11.7 μg/L (2.6, 50.1). Energy adjusted median dietary intakes of thiamine, riboflavin, niacin, vitamin B-6, folate, and vitamin B-12 were 0.85 mg/1000 kcal, 0.99 mg/1000 kcal, 8.6 mg/1000 kcal, 0.68 mg/1000 kcal, 219.76 μg/1000 kcal, and 1.67 μg/1000 kcal respectively. The median concentration of serum folate was 11.40 ng/ml (correlation with dietary folate intake: 0.07), and that of serum vitamin B-12 was 489.45 pg/ml (correlation with dietary vitamin B-12 intake: 0.12). Urinary %MMA had a median (range) of 9.8% (2.6%, 24.8%).

The median (range) of U-As concentrations and broad math and reading scores according to participant characteristics are shown in Table 1. Supplemental Table 1 presents a description of the Woodcock-Muñoz Achievement Battery subtests and the means (SD) and medians (range) of the W scores. Supplemental Table 2 presents participant characteristics in the complete case sample according to tertiles of broad reading scores (participant characteristics by tertiles of broad math scores are very similar and are not presented).

Table 2 shows the likelihood of transitioning from one math and reading score tertile to the next in relation to U-As, based on ordinal regression. U-As was not associated with the likelihood of moving across the tertiles of math or reading achievement scores. For broad math, there was some indication of an association, however, in the direction that was not hypothesized (OR=1.02; 95% CI: 1.00, 1.04; p=0.05); every higher unit of U-As was associated with 2% higher odds of being in the upper tertile of math scores. Findings were similar in the complete-case and imputed samples, with overlapping confidence intervals. There was no evidence of a non-linear relationship between U-As and achievement (result not shown).

The results of interaction tests are presented in Table 3. Because serum folate and vitamin B-12 concentrations could only be assessed among a subset of the complete case sample, the sample size for the interaction between U-As and serum folate was 222, and that for serum vitamin B-12 was 221. An interaction was observed between U-As and serum vitamin B-12 concentrations with broad math score in the complete case, OR = 1.12 (95% CI: 1.02, 1.24), as well as the imputed sample, OR = 1.05 (95% CI: 1.02, 1.09).

Discussion

We found little evidence that low level arsenic exposure affected children’s math or reading scores on the Woodcock-Muñoz Achievement Battery. In addition, there was no effect modification by urinary %MMA, the B-vitamin index, or dietary folate intake. Our exploratory analysis showed evidence of effect modification of the U-As-math achievement association by serum vitamin B-12; increased U-As and serum vitamin B-12 levels were associated with improved math performance. In a previous study, we found no effect of low-level arsenic exposure on general intellectual ability in these same children (Desai et al., 2018). In contrast, a recent study found an inverse association between low-level arsenic exposure and motor function in preschool children from Spain (Signes-Pastor et al., 2019), whereas several studies have shown inverse associations with intelligence and visual-spatial ability in geographical areas characterized by elevated arsenic exposure through drinking water (Nahar et al., 2014; Rosado et al., 2007; Wasserman et al., 2007; Wasserman et al., 2004). In Bangladesh, where arsenic exposure had been linked with intellectual ability deficits (Hamadani et al., 2011; Wasserman et al., 2007), arsenic in water was not related to achievement (Khan et al., 2012).

To our knowledge, the study from Bangladesh (Khan et al., 2012) represents the only other report on arsenic exposure and academic achievement in children. Our findings are consistent but few similarities exist between the two studies. First, arsenic exposure in Bangladesh was higher (water arsenic mean, range: Bangladesh, 119.5, 0.1–1263.2 μg/L vs. Uruguay, 0.63, 0.1–18.9 μg/L). Furthermore, academic achievement in Bangladesh was measured as annual scores on the national tests of English, Bangla, and math, whereas we assessed achievement using a standardized battery based on competencies children typically achieve by a certain age. Additionally, children from Bangladesh were older than children in our study (age 8–11 years in Bangladesh vs. 5–8 years in Uruguay). The scarcity of evidence, and the differences between the two studies underscore the need for further research on this topic. For example, behavior, social interactions, self-efficacy, parental involvement in schooling, and other factors within household and school predict school achievement; it would be important to understand how they contribute to achievement in the context of arsenic exposure.

The role of methylation capacity in arsenic neurotoxicity among children is not clear. Poor methylation capacity among Taiwanese children was a risk factor for developmental delays (Hsieh et al., 2014; Hsueh et al., 2016). The only two studies on arsenic and children’s neurodevelopment that included stratification by %MMA did not show evidence of effect modification (Desai et al., 2018; Hamadani et al., 2011). One of these studies was conducted in Montevideo by our group (Desai et al., 2018). It is of note that no standard definition of “high” or “low” %MMA exists, perhaps because various factors affect methylation of arsenic, including the consumption of methylated arsenic species directly from the diet. We did not observe any effect modification by urinary %MMA in the current analyses, which is perhaps unsurprising given our previous findings on cognitive performance (Desai et al., 2018).

Because B-vitamins participate in arsenic methylation, we hypothesized that B-vitamin intake/status would modify the arsenic-achievement relationship. However, our findings did not support this hypothesis, possibly because of the low variability of B-vitamin intake observed in this population. The Recommended Daily Allowance (RDA) for 4–8 year olds is 0.6 mg/day for thiamine, riboflavin and vitamin B-6, 0.8 mg/day of niacin equivalents, 200 μg/day of dietary folate equivalents, and 1.2 μg/day of vitamin B-12 (Institute of Medicine Standing Committee on the Scientific Evaluation of Dietary Reference et al., 1998). Commercially produced wheat flour in Uruguay has been fortified with folate since 2006. In our sample, 68% and 48% of participants reported consuming various types of bread and cookies, respectively. Additionally, 99% of participants had intakes above the RDA for thiamine, riboflavin, niacin, vitamin B-6, and vitamin B-12; 90% exceeded the RDA for folate. Thus, participants in our study appeared to have largely adequate levels of B-vitamin intake, another potential reason behind the observed results. However, determining sufficiency/deficiency of vitamin intake on the basis of 24-hour recalls may not be accurate, thus caution is advised when interpreting these findings. We observed interaction effects between serum vitamin B-12 and U-As on broad math scores, indicating that higher U-As as well as serum vitamin B-12 were associated with better math performance. Overall, because U-As and serum vitamin B-12 were associated with lower likelihood of moving to into higher math score categories, the positive interaction term represents a fairly small effect. It is also important to note that we were unable to further explore this interaction in stratified analysis because the proportional odds assumption was not met in the serum B-vitamin strata, and the sample sizes were further reduced.

Arsenobetaine is an organic arsenical commonly found in fish, and is excreted from the body without undergoing any chemical changes (Navas-Acien et al., 2011). On the other hand, arsenolipids and arsenosugars, also found in seafood, are metabolized to DMA (Raml et al., 2005; Schmeisser et al., 2006). Rice is another important source of inorganic arsenic as well as DMA (Schoof et al., 1999; Signes-Pastor et al., 2016; Williams et al., 2005). Higher urinary DMA could result from efficient methylation of arsenic, or from consumption of seafood and to some extent rice. As a result, urinary %MMA is regarded as a more reliable marker of arsenic methylation capacity than urinary %DMA (Vahter, 2001). In our study, only 20 participants reported any consumption of seafood and about half of the participants reported any rice intake. Seafood intake may lead to increased urinary DMA, thereby overestimating total urinary arsenic. We re-analyzed our data by excluding the 20 participants with any seafood intake; the results did not change (Supplemental Table 3). Because seafood intake may also be associated with neurodevelopmental outcomes due to omega fatty acids (Oken and Bellinger, 2008), we then adjusted our models for seafood intake; the results did not change (Supplemental Table 4). We did not adjust the main models for rice intake because arsenic exposure was measured as the sum of urinary inorganic arsenic, MMA, and DMA; adjusting for rice would mean adjusting for the exposure itself. The borderline statistically significant positive association between urinary arsenic exposure and broad math score that we observed could be due to chance. Further, while it is likely that study was underpowered, even after imputation, it does provide effect size estimates in the context of low-level arsenic exposure.

Our study has certain limitations. First, we had a participation rate of 53%, which could have resulted in selection bias if participation was associated with both the exposure and outcome. Arsenic exposure is not routinely tested in Uruguay, so it is very unlikely that participation decisions would be made on that basis; therefore, the likelihood of selection bias is low. Second, of the 357 children who participated in our study, urinary arsenic measures were obtained from 327 children and academic achievement was assessed among 310 children. Although exclusion of children with missing information resulted in a complete case sample of 239 participants, we conducted multiple imputations using chained equations to impute covariates among those with complete data on the outcomes of interest, thereby increasing our sample to 310 participants. Thus, of these 310 participants, 77% (n=239) had complete data, and 23% (n=71) had data imputed for one or more variables. The use of a robust imputation technique, and the fact that we observed similar results in the imputed and the complete case sample increase confidence in our findings. Third, dietary intake of B-vitamins was derived from two different databases, including the USDA food composition database. Although we matched as closely as possible the foods consumed by the participants to those listed in the USDA database, measurement error is possible. We did not specifically query B-vitamin supplement use, but did ask whether participants took iron supplements because they are more commonly used for children. Only 6% participants reported any intake of iron supplements, leading us to believe that overall supplement use in this population would be low. Nonetheless, misclassification of B-vitamin intake due to supplement use is possible.

The strengths of this study include the use of the standardized Woodcock-Muñoz Achievement Battery to assess academic achievement, which makes our findings comparable to others using this instrument (on the other hand, the use of country-specific assessments, while reflective of educational norms, makes comparison across studies challenging). Children are on their academic achievement trajectories by the end of their third grade and follow those until the end of schooling (Alexander et al., 1988). Our study captures an important part of this trajectory. In addition to the two 24-hour diet recalls, we had serum folate and vitamin B-12 levels, which is another strength. We adjusted our models for measures of lead and manganese exposure to account for their potential effects on the outcome, and HOME inventory scores to account for developmental inputs the child receives at home.

Conclusion

We found little evidence that low-level exposure to arsenic affects academic achievement among children attending first grade of school. We also found no impact of B-vitamin status, intake or methylation capacity on the relationship between urinary arsenic and achievement.

Supplementary Material

Acknowledgements

Abbreviations:

Footnotes

References