Abstract
Background
Our previous study demonstrated that prenatal manganese exposure is associated with cognitive test scores among a cohort of 2- to 3-year-old Bangladeshi children. This study tested the hypothesis that the adverse effects of manganese are mediated through poor prenatal growth.
Methods
Pregnant mothers were enrolled in a birth cohort in Bangladesh between 2008 and 2011, and children were followed at birth and age 20–40 months. Manganese concentration was measured in umbilical cord blood. Anthropometric measurements (weight, length, head circumference) were assessed at delivery. Children’s cognitive development was assessed at age 20–40 months using the Bayley Scales of Infant and Toddler Development—Third Edition. Using recently developed statistical approaches that estimate mediation and interaction effects simultaneously, we evaluated whether the association between cord blood manganese and cognitive score was mediated through anthropometric measures at birth.
Results
This analysis included 764 mother–child pairs. Higher manganese concentration was associated with lower cognitive score [β=–0.61, standard error (SE)=0.23, p = 0.009]. Among the birth measures, we found a significant indirect effect only through birth length (β =–0.10, SE = 0.03, p = 0.001). We also found evidence of mediated interaction (both mediation and interaction, β =–0.03, SE = 0.01, p = 0.01) with birth length in the association between cord blood manganese and cognitive score. The overall proportion mediated by birth length was 33% (p = 0.02) and the proportion attributed to interaction was 11% (p = 0.04). We did not find evidence of a mediating effect through birth weight or head circumference.
Conclusions
Our findings confirm that prenatal growth, particularly birth length, contributes to the overall effect of environmental manganese exposure on a child’s cognitive development.
Keywords: environmental health, public health, causal mediation
Key Messages
The newly developed causal mediation method of decomposing the effects of in utero environmental manganese exposure on cognitive development in early childhood can provide useful insights into the underlying mechanisms.
The adverse effects of manganese on early childhood neurodevelopment are mediated through poor prenatal growth, as shown by shorter birth length.
The effect of manganese on cognitive development differs across infants with varying birth lengths.
Introduction
Manganese is involved in carbohydrate, amino acid and lipid metabolism, as well as bone formation, all of which are crucial for proper growth, development and maintenance of the body over its life course.1,2 Dietary intake of manganese is considered essential to health; however, excess intake or exposure to manganese is associated with adverse neurological outcomes in children.3,4 In recent years, a growing body of evidence suggests that the benefit or toxicity of manganese is determined by the level of exposure.3
The primary source of manganese in the US population is through the diet,5 but many regions around the world have higher exposure to manganese through occupational and other environmental exposures, such as drinking water.6–8 In Bangladesh, where the current study takes place, exposure to high levels of environmental metals from drinking water is prevalent. Water extracted from shallow aquifers is the primary source of drinking and cooking water for most of the population of over 140 million.9 The majority of these shallow tube wells have levels of manganese above the World Health Organization guideline values.6,9 Prenatal exposure to environmental manganese, in particular, poses a threat in this population because manganese is actively transferred across the human placenta from mothers to newborns, and consequently increases the burden of manganese toxicity among both the mothers and their children.10 We and others have documented the association between environmental manganese exposure and poorer performance on cognitive assessments.3,4,11–18 Despite the growing evidence of manganese exposure as a neurotoxin, the mechanisms by which environmental manganese exposure impairs neurodevelopment are still not clearly defined.
Poorer performance on neurodevelopmental tests has been associated with lower anthropometric measurements at birth.12,19–22 Recently, we investigated the association between anthropometric measurements at birth and neurodevelopment status among 2- to 3-year-old Bangladeshi children. Higher physical birth parameters were associated with improved neurodevelopmental scores above and beyond the contribution of environmental metal exposure.23 This outcome was unexpected, given that the adverse effects of prenatal environmental manganese exposure on child growth12,24–29 and neurodevelopment are well established.3,4,11–18 To explain our results, we speculated that birth parameters may be mediators that link the relationship between in utero exposure to environmental metals and early childhood neurodevelopment.
Therefore, the primary purpose of this study was to test the hypothesis that the adverse effect of manganese on neurodevelopment is mediated through poor perinatal growth, as assessed by anthropometric measurements at birth including weight, length and head circumference. Potential modifying effects of anthropometric measures at birth were also assessed on the relationship between prenatal manganese exposure and cognitive outcome at 20–40 months of age. We used recently developed statistical approaches that allowed estimating mediation and interaction effects simultaneously.30,31
Methods
Study sample
Participants were members of an ongoing birth cohort study in Bangladesh. The design has been described previously.19,32 Briefly, pregnant women were recruited from clinics operated by Dhaka Community Hospital in the Sirajdikhan and Pabna Sadar Upazilas. When children were aged 12–40 months, we re-contacted parents and invited them and their children to participate in follow-up activities. This included two visits: one at 12 months and one at 20–40 months of age. The Human Research Committees at the Harvard T.H. Chan School of Public Health and Dhaka Community Hospital approved this study. All adult participants provided written informed consent and parents provided consent for their children. The Institutional Review Board at Boston Children’s Hospital formally ceded review of this study to the Harvard T.H. Chan School of Public Health.
Anthropometric measurements
Anthropometric measurements, including weight, length and head circumference, were assessed at the time of delivery by study staff. Birth weight was measured to the nearest 0.1 kg using a calibrated digital infant scale. Birth length was measured to the closest 0.1 cm with extended legs and heels against the measuring board using an infantometer. Head circumference was measured to the closest 0.1 cm at the maximal occipital-frontal circumference using a standard measuring tape.
Cognitive development
Cognitive developmental status was assessed at 20–40 months of age using a translated and culturally adapted version of the Bayley Scales of Infant and Toddler Development—Third Edition (BSID-III). Trained field staff administered the assessment using a standardized protocol. The quality of the assessments/scores was verified by a trained paediatrician and a paediatric neurologist.
Umbilical cord blood metal concentrations
Umbilical cord blood samples were collected at delivery. Detailed methods for the analysis of cord blood metal concentrations have been described previously.33 Briefly, metal concentrations were analysed by inductively coupled plasma mass spectrometry following an acid digestion method. Average limits of detection for manganese, arsenic and lead concentrations were 0.02 µg/dL, 0.03 µg/dL and 0.02 µg/dL, respectively. None of the samples reached concentrations below the limit of detection.
Covariates
During study visits, trained study staff administered questionnaires designed to collect information including maternal age, gestational age, maternal educational level, exposure to tobacco smoke and medical history. Gestational age was determined by first-trimester ultrasound. Maternal average daily protein intake was assessed by administering a food frequency questionnaire at the third-trimester visit.25,34,35 Protein intake was classified as a variable indicating daily protein consumption as low, medium or high. Child’s haematocrit level was measured at the 20–40-month visit. In addition, the Home Observation for Measurement of the Environment (HOME) Inventory was administered to the mothers to assess the stimulation/support available to a child in the home environment.36
Statistical analysis
All the variables were evaluated for normality and potential outliers. Cord blood metal concentrations were natural logarithmically transformed to normalize their skewed distributions. Independent t-tests for continuous variables and Chi-square tests for categorical variables were used to assess potential differences in characteristics between included and excluded participants. One-way analysis of variance (ANOVA) was used to compare the differences in children’s cognitive scores according to their age group. Child’s age- and sex-adjusted Pearson correlation coefficients were computed to test the simple correlation among children’s physical birth parameters.
We employed recently developed statistical approaches by Valeri and VanderWeele to estimate the proportion of the total effect of manganese that was mediated through growth parameters. In addition, this method allows estimating effect modification.30,31 Two multivariable-adjusted regression models were established (Figure 1). Outcome models included cord blood manganese and each anthropometric measure at birth, interaction term for cord blood manganese and each anthropometric measurement as exposures and cognitive score as an outcome. The mediator model included cord blood manganese as an exposure and each anthropometric measure as an outcome. All models were adjusted for covariates/confounders, including child’s age, child’s sex, gestational age, maternal age, maternal education, maternal protein intake, exposure to tobacco smoke, HOME score, child’s haematocrit level, study site, and cord blood arsenic and lead concentrations.
Figure 1.
Outcome and mediator models for the associations between cord blood manganese, physical birth parameters (birth weight, birth length and birth head circumference) and cognitive test scores measured at child’s age 20–40 months.
We further decomposed the total effect of manganese on neurodevelopment into four components: (i) effect due to mediation via anthropometric measures (pure indirect effect); (ii) effect due to interaction between manganese exposure and anthropometric measures (reference interaction); (iii) effect due to both mediation and interaction (mediated interaction); and (iv) effect due to neither mediation nor interaction (controlled direct effect). Detailed explanation is given in Table 1. The causal interpretation of the effect estimates relied on the assumption of correct model specification and the assumptions of no unmeasured confounding in the relationship between (i) the exposure and outcome, (ii) exposure and mediator and (iii) mediator and outcome. In addition, it is assumed that none of the mediator–outcome confounders is itself affected by the exposure. The four composite effects and the relationship between outcome and mediator regression models are described in Figure 1. In addition, estimates for the controlled direct effect were obtained from minimum to maximum values for each anthropometric measure at birth to explore the patterns of the association between cord blood manganese and cognitive score.
Table 1.
Definitions of the four-way decomposition of the total effect
| Effect | Counterfactual definitiona | Interpretation | Contextual definition |
|---|---|---|---|
| Total effect | (Ya – Ya*) | Total effect of exposure A (changing from a* to a) on the outcome Y | Overall effect |
| Pure indirect effect or mediated main effect | (Ya*m − Ya*m*) (Ma − Ma*) = (Ya*Ma − Ya*Ma*) | Effect of the mediator (changing from m* to m) on outcome Y when exposure A is a, multiplied by the effect of exposure A (changing from a* to a) on the mediator M | Due to mediation only |
| Reference interaction | (Yam − Yam* – Ya*m + Ya*m*) (Ma*) | An additive integration that operates only if the mediator is present (Ma* ≠ 0) when the exposure A is a | Due to interaction only |
| Mediated interaction | (Yam − Yam* – Ya*m + Ya*m*) (Ma − Ma*) | An additive interaction that operates only if the exposure A (changing from a* to a) has an effect on the mediator M (Ma − Ma* ≠ 0) | Due to mediation and interaction |
| Controlled direct effect | (Yam* − Ya*m*) | Effect of exposure A (changing from a* to a) on outcome Y intervening to fix the mediator M to m | Due neither to mediation nor interaction |
Y is the outcome, A is the exposure of interest and M is the potential mediator. For example, Yam is denoted as the value of the outcome Y that would have been observed when outcome A is set to level a and mediator M is set to level m.
We did not further adjust our analyses for multiple testing because the primary purpose of this investigation was hypothesis generation (i.e., to explore the mediating and modifying effects). A two-tailed p < 0.05 was considered significant. All analyses were performed using Stata/SE 13.1 software (StataCorp, College Station, TX, USA).
Results
Study population
A total of 764 mother–child pairs were included in our study. Among 825 mother–infant pairs who participated in neurodevelopmental assessment at 20–40 months, we excluded 61 pairs due to missing data on cord blood metals (n = 7), BSID-III cognitive score (n = 10), gestational age (n = 3), maternal protein intake (n = 2), exposure to tobacco smoke (n = 1), HOME score (n = 6) and haematocrit level (n = 31). A comparison of the characteristics of the study sample and those excluded or unavailable revealed no significant differences. Some exceptions were noted where lower gestational age, higher birth length and higher cognitive score were observed among included participants. The characteristics of study participants are shown in Table 2 and distributions of the cord blood metal concentrations are shown in Table 3. The median cord blood metal concentrations were 5.84 µg/dL for manganese, 0.58 µg/dL for arsenic and 3.15 µg/dL for lead. The summary of the cognitive scores according to age group is shown in Supplementary Table 1, available as Supplementary data at IJE online. Child’s age- and sex-adjusted Pearson correlation coefficients among physical birth parameters are shown in Supplementary Table 2, available as Supplementary data at IJE online. Although the physical birth parameters are moderately to strongly correlated (all p < 0.0001), only 9.0% [r2 = (0.30)2 =9.0%] and 9.6% [r2 = (0.31)2 = 9.6%] of the variation in birth length was explained by birth weight and birth head circumference, respectively.
Table 2.
Characteristics of study cohort
| Variables | Overall participants (N = 764) |
|---|---|
| Mothers | |
| Age at enrolment (years) | 23 (4.2) |
| Education (%) | |
| Primary or less | 47 (359) |
| Secondary or greater | 53 (405) |
| Exposure to tobacco smoke (%) | |
| Yes | 43 (326) |
| No | 57 (438) |
| Protein intake (%)a | |
| Low | 25 (193) |
| Medium | 51 (393) |
| High | 23 (178) |
| Study site (%) | |
| Sirajdikhan | 50 (383) |
| Pabna | 50 (381) |
| Children | |
| Gestational age (weeks) | 38 (2.0) |
| Age at 20- to 40-month visit (months) | 28 (2.9) |
| Female (%) | 50 (378) |
| Anthropometric measures at birth | |
| Weight (kg) | 2.9 (0.4) |
| Length (cm) | 47 (2.3) |
| Head circumference (cm) | 33 (1.4) |
| Haematocrit at 20- to 40-month visit (%) | 34 (3.2) |
| BSID-III cognitive score at 20- to 40-month visit | 60 (4.8) |
Data are shown as means (standard deviations) for continuous variables or percentages (counts) for categorical variables.
Protein intake was classified as a categorical variable indicating daily protein consumption as low, medium or high.
BSID-III, Bayley Scales of Infant and Toddler Development—Third Edition.
Table 3.
Distribution of cord blood metal concentrations among children
| Cord blood metals | Minimum | 25th percentile | Median | 75th percentile | Maximum |
|---|---|---|---|---|---|
| Manganese (µg/dL) | 1.24 | 4.27 | 5.84 | 9.50 | 303.19 |
| Arsenic (µg/dL) | 0.07 | 0.37 | 0.58 | 1.06 | 27.72 |
| Lead (µg/dL) | 0.28 | 1.58 | 3.15 | 6.65 | 79.18 |
Outcome and mediator regression models
Results of the multivariable-adjusted models (outcome model) for the associations between cord blood manganese, anthropometric measures at birth and cognitive score are shown in Supplementary Table 3, available as Supplementary data at IJE online. Notably, only the interaction term between cord blood manganese and birth length was statistically significant [β = 0.30, standard error (SE) = 0.08, p < 0.0001].
Supplementary Table 4, available as Supplementary data at IJE online, describes the associations between cord blood manganese and each anthropometric measure (mediator model). Among the measures, birth length was the only parameter associated with cord blood manganese concentration, with each additional unit increment in log-transformed cord blood manganese concentration associated with a 0.46-cm decrease in birth length (SE = 0.13, p < 0.0001). Supplementary Table 5, available as Supplementary data at IJE online, describes the association between cord blood manganese concentration and cognitive score based on the multivariable-adjusted linear regression models with and without adjustment for each anthropometric measurement at birth. When the model was additionally adjusted for birth length, the observed association substantially attenuated (β = –0.38, SE = 0.23, p = 0.10), suggesting that birth length may lie in the pathway between prenatal manganese exposure and cognitive development as a mediator.
A four-way decomposition of the total effect
The four-way decompositions of the total effect of cord blood manganese concentration on the BSID-III cognitive score are shown in Table 4. These estimates were computed fixing birth parameters at the mean values (m* value in equations indicated in Figure 1). The total effect of cord blood manganese concentration on the BSID-III cognitive score was significant with estimates ranging from –0.28 to –0.31 (all p ≤ 0.02, Table 4), confirming that, when the cord blood manganese concentrations increased from median to the 75th percentile level (from 5.84 to 9.50 µg/dL), the BSID-III cognitive score decreased accordingly.
Table 4.
Effect of anthropometric measurements on cognitive score due to mediation and interaction with cord blood manganese
| Anthropometric measures | Effects | Estimate (standard error) | p-value | Proportion attributed (standard error) | p-value |
|---|---|---|---|---|---|
| Birth weight | Total effect | –0.28 (0.12) | 0.02 | – | – |
| Pure indirect effect | –0.03 (0.02) | 0.11 | 0.12 (0.08) | 0.14 | |
| Reference interaction | 0.00 (0.005) | 0.99 | 0.0003 (0.02) | 0.99 | |
| Mediated interaction | –0.007 (0.005) | 0.22 | 0.02 (0.02) | 0.26 | |
| Controlled direct effect | –0.24 (0.11) | 0.03 | 0.86 (0.09) | <0.0001 | |
| Birth length | Total effect | –0.31 (0.12) | 0.007 | – | – |
| Pure indirect effect | –0.10 (0.03) | 0.001 | 0.33 (0.14) | 0.02 | |
| Reference interaction | 0.0003 (0.01) | 0.98 | 0.00 (0.04) | 0.98 | |
| Mediated interaction | –0.03 (0.01) | 0.01 | 0.11 (0.05) | 0.04 | |
| Controlled direct effect | –0.17 (0.11) | 0.12 | 0.56 (0.18) | 0.001 | |
| Birth head circumference | Total effect | –0.30 (0.12) | 0.009 | – | – |
| Pure indirect effect | –0.02 (0.01) | 0.15 | 0.07 (0.05) | 0.19 | |
| Reference interaction | 0.00 (0.02) | 0.99 | 0.00 (0.01) | 0.99 | |
| Mediated interaction | 0.0003 (0.004) | 0.99 | 0.00 (0.01) | 0.99 | |
| Controlled direct effect | –0.28 (0.12) | 0.02 | 0.93 (0.05) | <0.0001 |
Estimates were derived when the birth parameters were fixed at the mean values (birth weight = 2.9 kg, birth length = 47 cm, head circumference = 33 cm) and when cord blood manganese increased from median (= 5.84 µg/dL) to the 75th percentile (= 9.50 µg/dL).
Among the anthropometric measures at birth, we found a significant negative indirect effect (mediated main effect) through birth length (β =–0.10, SE = 0.03, p = 0.001). Moreover, we found evidence of mediated interaction (both mediation and interaction) (β = –0.03, SE = 0.01, p = 0.01). The controlled direct effect—the effect of the manganese concentration on cognitive score through pathways independently of anthropometric measures—was significant for those models that included birth weight or birth head circumference (all p ≤ 0.03), but not when length was considered in the model (estimate = –0.17, SE = 0.11, p = 0.12) (Table 4).
The overall proportion of the BSID-III cognitive score explained by the pure indirect effect (i.e. mediating effect) of birth length was 33% (p = 0.02), the proportion attributed to the effect of mediated interaction was 11% (p = 0.04) and the proportion explained solely by the effect of cord blood manganese concentration was 56% (p = 0.001) (Table 4). In addition, the overall proportion explained by the mediating effect of birth length (summation of the proportions due to mediated interaction and pure indirect effect) was 44% (p = 0.02).
Controlled direct effect of cord blood manganese on cognitive score
Controlled direct effects of cord blood manganese concentration on cognitive score, when anthropometric measures increased from minimum to maximum values, are shown in Figure 2. The figure depicts the trend of the association between cord blood manganese concentration and cognitive score when anthropometric measurements are fixed to specific values. In general, a linear trend was observed for the controlled direct effect. For birth weight, the increase in cord blood manganese from median to the 75th percentile was associated with a decrease in the BSID-III cognitive score only for those children with body weight 2.9 kg [estimate = –0.23, 95% confidence interval (CI) = –0.45, –0.002, p = 0.048] or lower (Figure 2a). For birth length, the direction of the controlled direct effect changed when the birth length reached 47.8 cm (estimate = –0.0004, 95% CI = –0.24, 0.24, p = 0.99). When the child’s birth length was 46.3 cm (estimate = –0.22, 95% CI = –0.44, –0.01, p = 0.04) or shorter, the increase in cord blood manganese concentration from median to the 75th percentile was significantly associated with lower BSID-III cognitive score. When birth length was 50.3 cm (estimate = 0.37, 95% CI = 0.005, 0.74, p = 0.047) or longer, the increase in cord blood manganese concentration from median to the 75th percentile was associated with higher BSID-III cognitive score (Figure 2b). For birth head circumference, the negative effect of manganese exposure on cognitive score was only seen in those children with birth head circumference ranging from 31.0 cm to 34.0 cm (Figure 2c).
Figure 2.
Controlled direct effect of cord blood manganese on cognitive scores when anthropometric measurements range from minimum to maximum values. (a) Birth weight, from minimum (0.8 kg) to maximum (4.5 kg) values. (b) Birth length, from minimum (34 cm) to maximum (64 cm) values. (c) Birth head circumference, from minimum (24 cm) to maximum (48 cm) values.
Discussion
Our previous studies have demonstrated that cord blood manganese concentration is associated with lower cognitive scores among 2- to 3-year-old children in Bangladesh.35 In this study, we employed new statistical techniques to examine causal pathways to test whether the effect of manganese on cognitive development was mediated through a reduction in physical growth during the prenatal period. Our principal findings are 2-fold. First, among the anthropometric measurements assessed at birth including weight, body length and head circumference, a mediating effect was observed only with birth length. Second, we found evidence of effect modification by birth length on the associations between cord blood manganese concentrations and cognitive score, suggesting that the effect of manganese on cognitive development differed across infants with varying birth lengths. Together, the newly developed method of decomposing the effects of in utero environmental manganese exposure on cognitive development in early childhood provided useful insights into the underlying mechanisms, which included both mediating and interacting effects of birth length.
Our study builds on a growing literature that has reported the associations between environmental manganese exposure with anthropometric measures at birth28,29 and adverse neurodevelopmental status,3,4,11,12,15–18 as well as the relationship between poor growth assessed by anthropometric measurement and lower neurodevelopmental test scores in early childhood.19–22 However, none of these prior studies assessed the linkage of these associations as part of a longitudinal mediation pathway starting with toxic chemical exposure in pregnancy. To our knowledge, this is the first study to assess causal mechanisms of the health effects of manganese through growth parameters.
Our a priori hypothesis was that all of the anthropometric measurements at birth are mediators linking the associations between prenatal manganese exposure and early childhood cognitive development. Unexpectedly, we found evidence of a mediating effect only through birth length, and not with birth weight or birth head circumference, even though these measurements are frequently highly correlated. Our child’s age- and sex-adjusted Pearson correlation coefficients confirmed that approximately 10% of the variation in birth length can be explained by birth weight or head circumference (Supplementary Table 2, available as Supplementary data at IJE online). Accordingly, factors that cause the variation in birth length that cannot be explained by birth weight and head circumference may be important contributors to the casual mechanism of the adverse effect of manganese on early childhood cognitive development.
The association between cord blood manganese concentration and birth length may be explained by the functional properties of manganese, where manganese is a primary component of the bone matrix and an important cofactor for enzymes necessary for bone metabolism.1,37 The manganese overexposure may have failed to fulfil its function and resulted in decreased fetal growth.38 Among the anthropometric measurements, birth length is more likely to be affected by the structure and metabolism of the bones, whereas head circumference is more closely associated with brain volume.39 Another explanation is the oxidative stress caused by high environmental manganese exposure, which may have led to impairment in cellular function and growth. Similar to iron, manganese is a transitional metal that catalyses oxidative cellular reactions.40 A prior study showed that iron, which shares similar chemical properties with manganese, was associated with poor physical birth outcome measures.41 Consequently, restriction in intrauterine growth, as demonstrated by shorter birth length, may have adversely affected the neurodevelopment in early childhood.
Potential mechanisms may also elaborate our findings regarding the effect modification by birth length. Shorter birth length may be viewed as an indication of intrauterine growth restriction that is affected by not only prenatal environmental manganese exposure, but also poor maternal and fetal nutrition.42 Accordingly, the interaction between manganese exposure and constraints in maternal and fetal nutrition (as manifested by shorter birth length) during pregnancy may have further restricted the cognitive development during the crucial period. Furthermore, it is plausible that the children born with shorter body length may have a reduced ability to maintain manganese homeostasis postnatally and have increased susceptibility to impaired neurodevelopment. This explanation supports the differences in the direction of the manganese effect on cognitive score (controlled direct effect) depending on the birth length. In particular, manganese functioned as a neurotoxin for children with birth length 46.3 cm or shorter, whereas manganese exposure improved the cognitive scores for those with birth length 50.3 cm or higher. Given that 50% (n = 384) of our study participants had birth length 46.3 cm or shorter and 2% (n = 18) of our participants had birth length 50.3 cm or higher, the neurotoxic effect of manganese is predominant and more of a concern in our study population.
The present study used anthropometric parameters assessed at birth (weight, length and head circumference) as indirect measures of prenatal growth. Future studies that incorporate changes in the growth curve during pregnancy may provide important data needed to better understand the relationship between prenatal exposure to environmental manganese, prenatal growth and cognitive development in early childhood.
The limitations of our study include the possibility that our findings may not be generalizable outside of Bangladesh, as our Bangladeshi population is uniquely exposed to high levels of arsenic and manganese through groundwater.19,25,32,43,26,44 The mean or the median cord blood manganese concentration observed in our study sample (Table 3) was higher than the values seen in other populations.3,28,29,45–47 These high environmental metal exposures may have resulted in significant findings of this study. Accordingly, our findings may not be applicable to those with lower environmental metal exposures. Although we adjusted our statistical models for arsenic and lead, we may not have captured the effect of multi-exposure to these environmental toxicants and fully encompassed the mediating effect with the anthropometrics measurements and the outcome of cognitive development. Furthermore, unmeasured confounding might affect our estimates. For example, sources of manganese exposure include food and water, access to which may be a correlate of better nutrition during pregnancy, particularly in a low-income country, such as Bangladesh. Better nutrition in pregnancy would likely predict higher cognitive scores.48,49 Other unmeasured confounders are also possible; however, due to the nature of the observational study, not all confounders can be measured and should be noted as a limitation. Moreover, our approach assumed a linear relationship between exposure, mediator and outcome and no mediator- or exposure-confounder interactions. Although the linear relationship for manganese and anthropometric measures were checked and validated, interaction with confounders might still be present.
Conclusions
Using newly developed statistical methods to evaluate causal mechanisms, we found evidence that the association between prenatal manganese exposure and reduced neurodevelopmental test scores in early childhood was mediated through the effects of manganese on birth length. This suggests that manganese may have effects on global nutritional status and growth that contribute to neurotoxicity, as opposed to effects that are specific to brain development. In addition, our analysis suggests that birth length also partly modifies the effect of manganese on neurodevelopment.
Supplementary Material
Acknowledgements
This work was supported by the United States National Institute of Environmental Health Sciences grants # R01 ES011622, R01 ES026317, ES P42 ES016454 and P30 ES000002. All authors have commented on the manuscript and read and approved the final version of the manuscript. The lead author will act as a guarantor for the paper. The references have been checked for accuracy and completeness.
Conflict of interest: The authors have no conflict of interest to declare.
References
- 1. Fraga CG. Relevance, essentiality and toxicity of trace elements in human health. Mol Aspects Med 2005;26:235–44. [DOI] [PubMed] [Google Scholar]
- 2. Trumbo P, Yates AA, Schlicker S, Poos M.. Dietary reference intakes: vitamin A, vitamin K, arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and zinc. J Am Diet Assoc 2001;101:294–301. [DOI] [PubMed] [Google Scholar]
- 3. Claus Henn B, Ettinger AS, Schwartz J. et al. Early postnatal blood manganese levels and children’s neurodevelopment. Epidemiology 2010;21:433–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Claus Henn B, Schnaas L, Ettinger AS. et al. Associations of early childhood manganese and lead coexposure with neurodevelopment. Environ Health Perspect 2012;120:126–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Dobson AW, Erikson KM, Aschner M.. Manganese neurotoxicity. Ann NY Acad Sci 2004;1012:115–28. [DOI] [PubMed] [Google Scholar]
- 6. Kinniburgh D, Smedley P.. Arsenic Contamination of Groundwater in Bangladesh: Final Report Summary Keyworth: Bangladesh Department for Public Health Engineering, British Geological Survey, 2001. http://www.bgs.ac.uk/arsenic (8 June 2017, date last accessed).
- 7. Li GJ, Zhang LL, Lu L, Wu P, Zheng W.. Occupational exposure to welding fume among welders: alterations of manganese, iron, zinc, copper, and lead in body fluids and the oxidative stress status. Int J Occup Environ Med 2004;46:241–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Frisbie SH, Mitchell EJ, Mastera LJ. et al. Public health strategies for Western Bangladesh that address arsenic, manganese, uranium, and other toxic elements in drinking water. Environ Health Perspect 2009;117:410–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Hasan S, Ali MA.. Occurence of manganese in groundwater of Bangladesh and its implications on safe water supply. J Civ Eng 2010;32:121–28. [Google Scholar]
- 10. Krachler M, Rossipal E, Micetic-Turk D.. Trace element transfer from the mother to the newborn—investigations on triplets of colostrum, maternal, and umbilical cord sera. Eur J Clin Nutr 1999;53:486–94. [DOI] [PubMed] [Google Scholar]
- 11. Riojas-Rodríguez H, Solís-Vivanco R, Schilmann A. et al. Intellectual function in Mexican children living in a mining area and environmentally exposed to manganese. Environ Health Perspect 2010;118:1465–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sanders AP, Henn BC, Wright RO.. Perinatal and childhood exposure to cadmium, manganese, and metal mixtures and effects on cognition and behavior: a review of recent literature. Curr Envir Health Rpt 2015;2:284–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Roels H, Bowler R, Kim Y. et al. Manganese exposure and cognitive deficits: a growing concern for manganese neurotoxicity. Neurotoxicology 2012;33:872–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Lin CC, Chen YC, Su FC. et al. In utero exposure to environmental lead and manganese and neurodevelopment at 2 years of age. Environ Res 2013;123:52–57. [DOI] [PubMed] [Google Scholar]
- 15. Wasserman GA, Liu X, Parvez F. et al. Water manganese exposure and children’s intellectual function in Araihazar, Bangladesh. Environ Health Perspect 2006;114:124–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Wasserman GA, Liu X, Parvez F. et al. Arsenic and manganese exposure and children’s intellectual function. Neurotoxicology 2011;32:450–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Peres TV, Eyng H, Lopes SC. et al. Developmental exposure to manganese induces lasting motor and cognitive impairment in rats. Neurotoxicology 2015;50:28–37. [DOI] [PubMed] [Google Scholar]
- 18. Peres TV, Schettinger MRC, Chen P. et al. Manganese-induced neurotoxicity: a review of its behavioral consequences and neuroprotective strategies. BMC Pharmacol Toxicol 2016;17:57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Rodrigues EG, Bellinger DC, Valeri L. et al. Neurodevelopmental outcomes among 2-to 3-year-old children in Bangladesh with elevated blood lead and exposure to arsenic and manganese in drinking water. Environ Health 2016;15:44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Richards M, Hardy R, Kuh D, Wadsworth ME.. Birth weight and cognitive function in the British 1946 birth cohort: longitudinal population based study. BMJ 2001;322:199–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Shenkin SD, Starr JM, Deary IJ.. Birth weight and cognitive ability in childhood: a systematic review. Psychol Bull 2004;130:989. [DOI] [PubMed] [Google Scholar]
- 22. Gale CR, O'Callaghan FJ, Bredow M, Martyn CN; the Avon Longitudinal Study of Parents and Children Study Team. The influence of head growth in fetal life, infancy, and childhood on intelligence at the ages of 4 and 8 years. Pediatrics 2006;118:1486–92. [DOI] [PubMed] [Google Scholar]
- 23. Lee JJ, Kapur K, Rodrigues EG. et al. Anthropometric measures at birth and early childhood are associated with neurodevelopmental outcomes among Bangladeshi children aged 2-3 years. Sci Total Environ 2017;607–608:475–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Rahman A, Vahter M, Smith AH. et al. Arsenic exposure during pregnancy and size at birth: a prospective cohort study in Bangladesh. Am J Epidemiol 2009;169:304–12. [DOI] [PubMed] [Google Scholar]
- 25. Gleason KM, Valeri L, Shankar A. et al. Stunting is associated with blood lead concentration among Bangladeshi children aged 2-3 years. Environ Health 2016;15:103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Kile ML, Cardenas A, Rodrigues E. et al. Estimating effects of arsenic exposure during pregnancy on perinatal outcomes in a Bangladeshi cohort. Epidemiology 2016;27:173–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Rahman MM, Ng JC, Naidu R.. Chronic exposure of arsenic via drinking water and its adverse health impacts on humans. Environ Geochem Health 2009;31:189–200. [DOI] [PubMed] [Google Scholar]
- 28. Zota AR, Ettinger AS, Bouchard M. et al. Maternal blood manganese levels and infant birth weight. Epidemiology 2009;20:367–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Guan H, Wang M, Li X. et al. Manganese concentrations in maternal and umbilical cord blood: related to birth size and environmental factors. Eur J Public Health 2014;24:150–57. [DOI] [PubMed] [Google Scholar]
- 30. VanderWeele TJ. A unification of mediation and interaction: a 4-way decomposition. Epidemiology 2014;25:749–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Valeri L, VanderWeele TJ.. Mediation analysis allowing for exposure–mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods 2013;18:137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Kile ML, Rodrigues EG, Mazumdar M. et al. A prospective cohort study of the association between drinking water arsenic exposure and self-reported maternal health symptoms during pregnancy in Bangladesh. Environ Health 2014;13:29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Rodrigues EG, Kile M, Dobson C. et al. Maternal-infant biomarkers of prenatal exposure to arsenic and manganese. J Expo Sci Environ Epidemiol 2015;25:639–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Lin PI, Bromage S, Mostofa MG. et al. Validation of a dish-based semiquantitative food questionnaire in rural Bangladesh. Nutrients 2017;9:49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Valeri L, Mazumdar MM, Bobb JF. et al. The joint effect of prenatal exposure to metal mixtures on neurodevelopmental outcomes at 20-40 months of age: evidence from rural Bangladesh. Environ Health Perspect 2017;125:067015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Black MM, Baqui AH, Zaman K. et al. Iron and zinc supplementation promote motor development and exploratory behavior among Bangladeshi infants. Am J Clin Nutr 2004;80:903–10. [DOI] [PubMed] [Google Scholar]
- 37. Palacios C. The role of nutrients in bone health, from A to Z. Crit Rev Food Sci Nutr 2006;46:621–28. [DOI] [PubMed] [Google Scholar]
- 38. Vigeh M, Yokoyama K, Ramezanzadeh F. et al. Blood manganese concentrations and intrauterine growth restriction. Reprod Toxicol 2008;25:219–23. [DOI] [PubMed] [Google Scholar]
- 39. Bartholomeusz H, Courchesne E, Karns C.. Relationship between head circumference and brain volume in healthy normal toddlers, children, and adults. Neuropediatrics 2002;33:239–41. [DOI] [PubMed] [Google Scholar]
- 40. Hamai D, Bondy SC.. Oxidative basis of manganese neurotoxicity. Ann NY Acad Sci 2004;1012:129–41. [DOI] [PubMed] [Google Scholar]
- 41. Fellman V, Rapola J, Pihko H, Varilo T, Raivio KO.. Iron-overload disease in infants involving fetal growth retardation, lactic acidosis, liver haemosiderosis, and aminoaciduria. Lancet 1998;351:490–93. [DOI] [PubMed] [Google Scholar]
- 42. Walker SP, Wachs TD, Gardner JM. et al. Child development: risk factors for adverse outcomes in developing countries. Lancet 2007;369:145–57. [DOI] [PubMed] [Google Scholar]
- 43. Mazumdar M. Does arsenic increase the risk of neural tube defects among a highly exposed population? A new case–control study in Bangladesh. Birth Defects Res 2017;109:92–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Mazumdar M, Hasan MOSI, Hamid R. et al. Arsenic is associated with reduced effect of folic acid in myelomeningocele prevention: a case control study in Bangladesh. Environ Health 2015;14:1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Takser L, Mergler D, de Grosbois S, Smargiassi A, Lafond J.. Blood manganese content at birth and cord serum prolactin levels. Neurotoxicol Teratol 2004;26:811–15. [DOI] [PubMed] [Google Scholar]
- 46. Osada H, Watanabe Y, Nishimura Y, Yukawa M, Seki K, Sekiya S.. Profile of trace element concentrations in the feto‐placental unit in relation to fetal growth. Acta Obstet Gynecol Scand 2002;81:931–37. [DOI] [PubMed] [Google Scholar]
- 47. Rudge CV, Röllin HB, Nogueira CM, Thomassen Y, Rudge MC, Odland JØ.. The placenta as a barrier for toxic and essential elements in paired maternal and cord blood samples of South African delivering women. J Environ Monit 2009;11:1322–30. [DOI] [PubMed] [Google Scholar]
- 48. Veena SR, Krishnaveni GV, Srinivasan K. et al. Higher maternal plasma folate but not vitamin B-12 concentrations during pregnancy are associated with better cognitive function scores in 9-to 10-year-old children in South India. J Nutr 2010;140:1014–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Nyaradi A, Li J, Hickling S, Foster J, Oddy WH.. The role of nutrition in children’s neurocognitive development, from pregnancy through childhood. Front Hum Neurosci 2013;7:97. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.