AJCN. First published ahead of print January 21, 2015 as doi: 10.3945/ajcn.114.100503.

Prenatal exposure to methyl mercury from fish consumption and polyunsaturated fatty acids: associations with child development at 20 mo of age in an observational study in the Republic of Seychelles1–4 JJ Strain, Alison J Yeates, Edwin van Wijngaarden, Sally W Thurston, Maria S Mulhern, Emeir M McSorley, Gene E Watson, Tanzy M Love, Tristram H Smith, Kelley Yost, Donald Harrington, Conrad F Shamlaye, Juliette Henderson, Gary J Myers, and Philip W Davidson ABSTRACT Background: Fish is a rich source of n–3 polyunsaturated fatty acids (PUFAs) but also contains the neurotoxicant methyl mercury (MeHg). PUFAs may modify the relation between prenatal MeHg exposure and child development either directly by enhancing neurodevelopment or indirectly through the inflammatory milieu. Objective: The objective was to investigate the associations of prenatal MeHg exposure and maternal PUFA status with child development at 20 mo of age. Design: The Seychelles Child Development Study Nutrition Cohort 2 is an observational study in the Republic of Seychelles, a high fish-eating population. Mothers were enrolled during pregnancy and their children evaluated at 20 mo of age by using the Bayley Scales of Infant Development II (BSID-II), the MacArthur Bates Communicative Development Inventories (CDI), and the Infant Behavior Questionnaire–Revised. There were 1265 mother-child pairs with complete data. Results: Prenatal MeHg exposure had no direct associations with neurodevelopmental outcomes. Significant interactions were found between MeHg and PUFAs on the Psychomotor Developmental Index (PDI) of the BSID-II. Increasing MeHg was associated with lower PDI but only in children of mothers with higher n–6/n–3. Among mothers with higher n–3 PUFAs, increasing MeHg was associated with improved PDI. Higher maternal docosahexaenoic acid (DHA) was associated with improved CDI total gestures (language development) but was significantly adversely associated with the Mental Development Index (MDI), both with and without MeHg adjustment. Higher n–6/n–3 ratios were associated with poorer scores on all 3 CDI outcomes. Conclusions: We found no overall adverse association between prenatal MeHg exposure and neurodevelopmental outcomes. However, maternal PUFA status as a putative marker of the inflammatory milieu appeared to modify the associations of prenatal MeHg exposure with the PDI. Increasing DHA status was positively associated with language development yet negatively associated with the MDI. These findings may indicate existence of an optimal DHA balance with respect to arachidonic acid for different aspects of neurodevelopment. Am J Clin Nutr doi: 10.3945/ajcn.114.100503. Keywords child development, maternal fish consumption, methyl mercury, polyunsaturated fatty acids, docosahexaenoic acid, arachidonic acid, n–6/n–3 ratio, Mental Developmental Index, Psychomotor Developmental Index, language development

INTRODUCTION

The risks and benefits of fish consumption during pregnancy have received much attention in recent years (1–4). Studies have predominantly focused on methyl mercury (MeHg)5 exposure as a potential risk from fish consumption and on the n–3 PUFA, DHA, as a potential benefit. Although prenatal MeHg is a welldocumented neurotoxicant at high doses, the extent to which low-level exposure from fish consumption affects child development remains controversial. Results from prospective mother-child cohorts in the Republic of Seychelles have consistently shown no adverse associations between prenatal MeHg exposure and children’s subsequent development (5–11). Other large studies in the United Kingdom and Spain have reported similar findings (12, 13), whereas studies from New Zealand, the Faroe Islands, and the United States have reported adverse developmental influences of prenatal MeHg exposure (14–17). These inconsistencies may be related to variability in study designs, populations, genetic susceptibility, biomarkers of exposure or coexposures, or other factors. Alternatively, the benefits of fish 1 From the Northern Ireland Centre for Food and Health (NICHE), School of Biomedical Sciences, University of Ulster (JJS, AJY, MSM, and EMM); the School of Medicine and Dentistry, University of Rochester, Rochester, NY (EvW, SWT, GEW, TML, THS, KY, DH, GJM, and PWD); and the Child Development Centre, Ministry of Health, Mahe´, Republic of Seychelles (CFS and JH). 2 Supported by the NIH (grants R01-ES010219 and P30-ES01247) and inkind support from the government of Seychelles. 3 The study sponsors had no role in the design, collection, analysis, or interpretation of data; in the writing of the report; or in the decision to submit the article for publication. 4 Address correspondence to JJ Strain, Northern Ireland Centre for Food and Health (NICHE), School of Biomedical Sciences, University of Ulster, Cromore Road, Coleraine, Northern Ireland, BT52 1SA, United Kingdom. E-mail: [email protected]. 5 Abbreviations used: AA, arachidonic acid; BSID-II, Bayley Scales of Infant Development II; CDI, Communicative Development Inventory; IBQ-R, Infant Behavior Questionnaire–Revised; KBIT, Kaufman Brief Intelligence Test; MDI, Mental Developmental Index; MeHg, methyl mercury; NC1, Nutrition Cohort 1; NC2, Nutrition Cohort 2; PDI, Psychomotor Developmental Index; PROCESS, Pediatric Review of Children’s Environmental Support and Stimulation. Received September 30, 2014. Accepted for publication December 19, 2014. doi: 10.3945/ajcn.114.100503.

Am J Clin Nutr doi: 10.3945/ajcn.114.100503. Printed in USA. Ó 2015 American Society for Nutrition

Copyright (C) 2015 by the American Society for Nutrition

1 of 8

2 of 8

STRAIN ET AL.

consumption may outweigh or mask any potential adverse effects of MeHg on developmental outcomes (8). In a smaller previous cohort [Nutrition Cohort 1 (NC1)] we reported improved psychomotor performance with better prenatal status of n–3 PUFA at 9 and 30 mo (8, 9, 18) and improved verbal skills and comprehension at 5 y of age (11). In that cohort, children exposed to higher prenatal n–6 PUFA had poorer outcomes. However, in that study, there was no evidence to support the hypothesis that higher prenatal MeHg exposure was associated with adverse outcomes over 5 y of follow-up (11). On the basis of those findings, we hypothesized that the n–3 PUFA and other nutritional components associated with fish consumption might mask an association with prenatal MeHg exposure or that the nutritional benefits might exceed possible neurotoxicity. The toxic effects of MeHg on the developing brain are considered to be mediated by oxidative damage, which in turn causes inflammation (19). As both n–6 and n–3 PUFAs compete for the same enzymes in biosynthetic pathways and incorporation into cell membranes, the relative amounts of these PUFAs available in the diet are important for determining the physiologic n–6/n–3 balance. Furthermore, a high n–6/n–3 ratio in favor of n–6 PUFAs, which are more proinflammatory than n–3 PUFAs, may augment any possible inflammatory insults that might result from MeHg exposure in the brain (20). Accordingly, we enrolled a larger cohort (Nutrition Cohort 2; NC2) to confirm our earlier findings of no adverse effects of prenatal MeHg exposure and to clarify the role of prenatal PUFA status in either enhancing neurodevelopment and/or modifying any association with MeHg.

of the data and included independent data verification, data management, and statistical analysis. All investigators and staff other than those measuring MeHg exposure or directly preparing for and carrying out statistical analyses were blinded to MeHg data. Investigators measuring MeHg exposure and PUFA status were blinded to developmental outcome data. Statistical analysis plans based on biological hypotheses were developed a priori and are described in the statistical analysis section. Blood sampling and PUFA analysis At 28 wk of gestation, nonfasting maternal blood samples were collected. Samples were processed promptly and stored at 2808C until analyzed, as previously described (8). Blood samples were shipped at 2808C to the University of Ulster for serum total PUFA analysis, which was undertaken according to an adaptation of the method by Folch et al. (21). Fatty acid methyl esters were detected and quantified by using the gold-standard technique of gas chromatography–mass spectrometry (7890A-5975C; Agilent) using heptadecanoic acid (C17:0) as the internal standard, as previously described (8). All analytic standards were of $99% purity and purchased from Sigma-Aldrich. Total serum PUFA status was chosen as a biomarker to encompass recent PUFA concentrations of the triacylglycerol fraction, to which the majority of circulating PUFAs are bound during pregnancy and circulate to the fetus (22). We measured the individual PUFAs— linoleic acid, arachidonic acid (AA), a-linolenic acid, EPA, and DHA—and results were presented as milligrams per milliliter to indicate physiologic quantities. Methylmercury exposure

SUBJECTS AND METHODS

Study design and characteristics NC2 is part of the Seychelles Child Development Study, a multicohort observational study with the overall aim of investigating associations between prenatal MeHg exposure and child development. NC2 was conducted between 2008 and 2011 on Mahe´, the main island of the Republic of Seychelles. Mothers were recruited during their first antenatal visit (from 14 wk of gestation) at 8 health centers between January 2008 and January 2011. Power calculations were based on the magnitudes of associations observed for the Bayley Scales of Infant Development II (BSID-II) Psychomotor Developmental Index (PDI) at 9 and 30 mo in our earlier NC1 cohort (8, 18) by using a 2-sided significance level of a = 0.05; power for interaction models dichotomized PUFA at the median. We determined that a cohort of 1200 mother-child pairs would be sufficient to detect an interaction between n–6 PUFA and MeHg on the 30-mo PDI with 81% power, between n–3 PUFA and MeHg on the 9-mo PDI with 99% power, and between the n–6/n–3 ratio and MeHg on the 9-mo PDI with 99% power. A cohort size of 1200 would also be sufficient to detect main effects of n–3 and the n–6/n–3 ratio on both the 9-mo and the 30-mo PDI with 99% power. Inclusion criteria for NC2 included being native Seychellois, being $16 y of age, having a singleton pregnancy, and having no obvious health concerns. The study was reviewed and approved by the Seychelles Ethics Board and the Research Subjects Review Board at the University of Rochester. We used quality control procedures at all collaborating study sites to ensure the integrity

At delivery, maternal hair samples were collected to determine prenatal MeHg exposure. Total mercury was measured by the standard technique of atomic absorption spectroscopy at the University of Rochester in the longest hair segment available to reflect exposure throughout pregnancy. Hair was assumed to grow at a rate of 1.1 cm/mo (7). Mercury deposited in hair is w80% MeHg and is known to correlate with mercury deposited in the infant brain (7). Therefore, we refer to this measurement as the prenatal MeHg exposure. Developmental assessment When infants were about 20 mo of age (range: 15.9–28.4 mo), they completed the BSID-II, a well-established measure of cognition and development previously administered to previous Seychelles Child Development Study cohorts. The BSID-II Mental Developmental Index (MDI) and PDI are scaled scores obtained by direct examination of the child. Testing was implemented by specially trained nurses at the Child Development Centre, Mahe´. All study forms were shipped to the University of Rochester and double entered. Data from the 2 BSID-II endpoints were scaled according to child age at testing. Interobserver reliability for the BSID-II was determined for about 10% of the cohort by comparing the independent test scoring of 2 nurses. The median agreement on scored MDI and PDI items was 100%. At testing, mothers completed questionnaires on the Infant Behavior Record–Revised (IBQ-R), a measure of infant and toddler social and play behaviors, and the MacArthur-Bates

PRENATAL MeHg, PUFA, AND CHILD DEVELOPMENT

Communicative Development Inventories (CDI), a measure of social communication and early language development. Children were older than the normative group for the IBQ-R, but some were too young for the toddler version of this instrument, so we administered the IBQ-R and checked each outcome for ceiling effects. Similarly, many children were older than the normative group for the CDI when tested, so we analyzed only the 3 outcomes that did not show ceiling effects: total gestures, vocabulary produced, and vocabulary understood. We used raw scores for the 3 IBQ-R subscales: surgency, negative affect, and effortful control. We also used raw scores for CDI analysis as have other authors (23). For vocabulary produced and vocabulary understood, raw scores were square root transformed to meet regression model assumptions. Statistical analysis We calculated measures of central tendency and variability to describe demographic, exposure, nutritional, and developmental characteristics of mothers and children. We computed Pearson correlations between prenatal MeHg exposure and PUFA status and carried out linear regression to evaluate the main and interactive effects of MeHg and PUFA on outcomes. In main effects models, we evaluated DHA and AA because these PUFAs are considered to have a direct influence on brain development (20). In the MeHg by PUFA interaction models, we used total n–3, total n–6, and the n–6/n–3 ratio because the balance of these PUFAs can influence the inflammatory response to MeHg toxicity in the developing brain (24), and in turn might modify MeHg toxicity. We evaluated the n–6/n–3 ratio in models both with and without an interaction with MeHg. All models were fit by using the statistical package R version 3.0.2 (www.r-project.org). We examined the main effects of MeHg and PUFA on developmental outcomes with and without adjustment for each other. We then examined interactions between MeHg and tertiles of total n–3 and total n–6 PUFA and, in a separate model, interactions between MeHg and tertiles of the n–6/n–3 ratio. The main effects models used PUFA as continuous variables, whereas we used PUFA tertiles in the interaction model for interpretability, particularly to compare MeHg slopes among subjects with low, medium, or high PUFA levels. Model assumptions were checked by using standard methods (25), which included checking for linearity and constant variance and normality of the residuals. All models were adjusted for covariates known to be associated with child development (7)—namely, maternal age, child age at testing, child sex, Hollingshead socioeconomic status, and number of parents living with the child (family status). In secondary regression models, we also adjusted for mother’s cognitive ability [Kaufman Brief Intelligence Test (KBIT)] and the child’s home environment [Pediatric Review of Children’s Environmental Support and Stimulation (PROCESS)]. These data were available on only a subset of mothers (n = 1155 for KBIT and n = 1070 for PROCESS). To evaluate whether differences in MeHg and PUFA effects between the primary and secondary models resulted from adjustment for KBIT and PROCESS or from the different sample sizes, we also fit models by using the smaller set of observations without adjusting for KBIT and PROCESS. We used 2-sided a = 0.05 to determine statistical significance. The analyses were specified in an a priori analysis plan, developed before model fitting.

3 of 8

RESULTS

We recruited a total of 1535 mother-child pairs and conducted primary analyses on 1265 with complete covariate data after exclusions and a measure of at least one outcome (Figure 1). Summary statistics for selected maternal characteristics, prenatal MeHg and PUFA status, child outcomes, and model covariates in the cohort analyzed (n = 1265) are presented in Table 1. Mothers reported consuming an average of 8.5 fish meals per week during pregnancy, as assessed by a Fish Use Questionnaire. Pearson correlation analysis showed prenatal MeHg exposure to be positively correlated with serum concentrations of several n–3 PUFAs: a-linolenic acid (r = 0.07, P = 0.01), EPA (r = 0.08, P , 0.01), and DHA (r = 0.11, P , 0.01). The main effects associations between MeHg and PUFA with BSID-II outcomes are presented in Table 2, and those with CDI and IBQ-R outcomes are shown in Table 3. Results from interaction models are presented in Table 4.

MeHg associations Prenatal MeHg exposure both with and without adjustment for PUFA was not associated with any test score (Tables 2 and 3). In models that included total n–3 and total n–6 and did not adjust for KBIT and PROCESS, there were no statistically significant interactions between MeHg and n–6 PUFA tertiles for any outcome. Therefore, we report the results from these models with MeHg and n–3 PUFA tertile interactions only (Table 4). For the MDI, the interactions between MeHg and n–3 PUFA were not significant (P = 0.47; 2 df test). However, for the PDI, there was a significant MeHg by n–3 interaction (P , 0.01), indicating that the MeHg effect differed across tertiles of n–3 PUFA (Figure 2A). For subjects in the low and medium n–3 PUFA tertiles, increased MeHg exposure was not significantly associated with lower PDI scores. However, among subjects in the highest tertile of n–3 PUFA, the estimated MeHg slope showed a significant improvement in PDI scores with increasing MeHg exposure (P , 0.01). The MeHg by n–3 PUFA tertile interactions were also significant (P = 0.05) in secondary models adjusted for KBIT and PROCESS with similar MeHg slopes (data not shown).

FIGURE 1. Description of exclusions and missing data for the current analysis. BSID-II, Bayley Scales of Infant Development II; CDI, Communicative Development Inventories; IBQ-R, Infant Behavior Questionnaire– Revised; MeHg, methyl mercury.

4 of 8

STRAIN ET AL.

TABLE 1 Summary statistics for selected maternal characteristics, infant cognitive outcomes at 20 mo of age, and model covariates1 Variable Mothers Maternal age at enrollment, y Estimated weekly fish meals Hair MeHg, ppm Serum PUFA, mg/mL Linoleic acid AA a-Linolenic acid EPA DHA Total n–3 PUFA Total n–6 PUFA n–6/n–3 ratio AA/DHA ratio Developmental outcomes BSID-II MDI, scaled score BSID-II PDI, scaled score CDI total gestures CDI vocabulary produced CDI vocabulary understood IBQ-R surgency IBQ negative affect IBQ effortful control Covariates Child age at testing, mo Hollingshead SES Family status KBIT scaled PROCESS

n

Mean 6 SD

1265

26.85 6 6.33

16.03

46.56

1208

8.52 6 4.56

0.00

37.00

1265

3.92 6 3.46

0.00

31.66

1265 1265 1265 1265 1265 1265 1265 1265 1265

0.90 0.20 0.04 0.05 0.19 0.27 1.11 4.36 1.26

6 6 6 6 6 6 6 6 6

0.31 0.04 0.00 0.05 0.04 0.12 0.43 1.56 0.13

2.34 0.38 0.11 0.12 0.52 0.64 2.71 15.81 7.93

(P = 0.27) (Figure 3A), nor were there any significant interactions between MeHg and PUFA for any of the CDI and IBQ-R outcomes (data not shown).

Minimum Maximum

PUFA associations

0.25 0.08 0.01 0.01 0.09 0.09 0.30 1.63 0.77

1243

87.88 6 10.62

49.00

118.00

1241

96.90 6 10.40

49.00

125.00

1265 46.46 6 8.36 1265 123.60 6 87.60

2.00 0.00

63.00 388.00

1265 233.39 6 90.93

0.00

396.00

In main effect models, DHA was significantly adversely associated with the MDI score both with and without MeHg adjustment (Table 2). In the primary model adjusting for MeHg, with each 0.1-mg/mL serum increase in DHA, the MDI score was estimated to decline by 0.97 points. The n–6/n–3 ratio was significantly associated with an improved MDI score with or without adjustment for MeHg (Table 2). No significant associations between PUFA and the PDI score were observed. Higher DHA was associated with improved total gestures (P , 0.01) (Figure 3B). Higher n–6/n–3 ratios were associated with poorer scores on all 3 CDI outcomes of vocabulary produced (P = 0.02), vocabulary understood (P , 0.01), and total gestures (P , 0.01) (Table 3, Figure 3C). IBQ-R scores were not significantly associated with PUFA status (Table 3). Covariate associations Child sex and age and Hollingshead socioeconomic status were associated with MDI scores, and maternal age, child sex and age, and family status were associated with PDI scores (Table 2). DISCUSSION

1264 1264 1264

5.38 6 0.75 4.12 6 1.00 5.18 6 0.74

2.15 1.00 1.00

7.00 7.00 7.00

1265

20.71 6 1.34

15.90

28.39

6 6 6 6

11.00 0.00 40.00 96.00

63.00 1.00 126.00 200.00

1265 32.01 1265 0.74 1155 87.08 1070 156.41

10.31 0.44 17.03 16.47

1

Total n–3 PUFA = sum of ALA, EPA, + DHA; total n–6 PUFA = sum of LA + AA. AA, arachidonic acid; BSID-II, Bayley Scales of Infant Development II; CDI, Communicative Development Inventories; IBQ-R, Infant Behavior Questionnaire–Revised; KBIT, Kaufman Brief Intelligence Test; MDI, Mental Developmental Index; MeHg, methyl mercury; PDI, Psychomotor Developmental Index; PROCESS, Pediatric Review of Children’s Environmental Support and Stimulation; SES, socioeconomic status.

In primary interaction models of n–6/n–3 ratio tertiles, the interactions between MeHg and n–6/n–3 were not significant for the MDI (P = 0.93; 2 df test) (Table 4). For the PDI, however, there were significant MeHg by n–6/n–3 interactions (P = 0.02), indicating that the MeHg direct association differed across n–6/ n–3 ratio tertiles (Figure 2B). However, increased MeHg exposure was significantly associated with lower scores among subjects in the highest n–6/n–3 ratio tertile only (P = 0.03), indicating an adverse MeHg association. The estimated MeHg slopes for MDI and PDI within n–6/n–3 ratio tertiles were similar in secondary models adjusted for KBIT and PROCESS (data not shown). There were no significant direct associations between MeHg and any of the CDI outcomes (Table 3), including total gestures

The primary finding from this longitudinal observational study in the Seychelles was the absence of an overall association between MeHg exposure and child developmental outcomes at 20 mo of age. This finding did not change after adjusting for maternal PUFA status and confirms our findings from the NC1 and main cohorts, in which MeHg exposure had no consistent direct influence on cognitive outcomes during 5 and 19 y of follow-up, respectively (8, 26). Some studies have shown conflicting adverse neurotoxic effects of prenatal MeHg exposure (14–17), but such findings cannot be directly compared with the current study because of wide variability in study design, population, and choice of biomarkers of exposure or coexposures. The current cohort was large enough to test the extent to which MeHg associations with outcomes are modified by PUFA. There were 2 significant associations with the PDI: an adverse association of MeHg on the PDI at a high serum n–6/n–3 ratio and a beneficial association with MeHg at high n–3 status. These findings suggest the n–6 to n–3 PUFA balance is important to consider when studying MeHg associations at these levels of exposure and may reflect the capability of n–6 PUFA or n–3 PUFA, at higher concentrations, to augment or counteract, respectively, MeHg-induced inflammation. The n–6/n–3 ratio can be regarded as an indirect measure of inflammation, reflecting the potential for greater production of n–6 PUFA-derived eicosanoids, which are more proinflammatory than those derived from n–3 PUFA. The physiologic effects of a higher n–6/n–3 ratio have been associated with increased systemic inflammation and increased risk of disease (24). Other prospective studies have reported adverse associations of a high maternal n–6/n–3 ratio (high n–6 PUFA) with child development, including attention problems at age 5–6 y (27) and language at age 2 y (28). Another cross-sectional study

5 of 8

PRENATAL MeHg, PUFA, AND CHILD DEVELOPMENT TABLE 2 Main effects models for prenatal MeHg exposure and PUFA status variables and the BSID-II MDI and PDI at 20 mo of age1 20 mo MDI MeHg + PUFA (n = 1243) b DHA + AA MeHg DHA AA Maternal age Hollingshead SES Child sex (girl) Child age Family status n–6/n–3 ratio MeHg n–6/n–3 ratio Maternal age Hollingshead SES Child sex (girl) Child age Family status

SE

P

20 mo PDI

PUFA only (n = 1243) b

SE

MeHg only (n = 1243) P

b

SE

MeHg + PUFA (n = 1241) P

b

SE

PUFA only (n = 1241) b

P

SE

MeHg only (n = 1241) b

P

SE

P

20.06 29.73 8.13 20.07 0.12 2.98 21.57 0.76

0.08 0.46 20.08 0.08 0.31 0.02 0.09 0.80 0.03 0.09 0.68 4.33 0.02* 210.11 4.30 0.02* 5.54 4.4 0.21 5.67 4.37 0.19 4.47 0.07 8.34 4.46 0.06 22.4 4.54 0.60 22.48 4.53 0.58 0.05 0.17 20.07 0.05 0.15 20.08 0.05 0.10 20.12 0.05 0.01* 20.12 0.05 0.02* 20.11 0.05 0.02* 0.03 ,0.01* 0.12 0.03 ,0.01* 0.13 0.03 ,0.01* 0.04 0.03 0.19 0.04 0.03 0.19 0.03 0.03 0.24 0.58 ,0.01* 2.99 0.58 ,0.01* 2.95 0.58 ,0.01* 1.67 0.59 ,0.01* 1.67 0.59 ,0.01* 1.69 0.59 ,0.01* 0.23 ,0.01* 21.56 0.22 ,0.01* 21.59 0.22 ,0.01* 20.49 0.23 0.03* 20.49 0.23 0.03* 20.45 0.22 0.04* 0.68 0.26 0.77 0.68 0.26 0.81 0.68 0.23 21.4 0.69 0.04* 21.4 0.69 0.04* 21.43 0.69 0.04*

20.07 0.38 20.07 0.12 2.97 21.53 0.74

0.08 0.39 0.18 0.04* 0.05 0.13 0.03 ,0.01* 0.58 ,0.01* 0.22 ,0.01* 0.68 0.27

0.39 20.08 0.12 2.98 21.51 0.75

20.08 0.08 0.31 0.03 0.09 0.70 0.03 0.09 0.68 0.18 0.03* 20.07 0.18 0.72 20.07 0.18 0.70 0.05 0.11 20.08 0.05 0.10 20.11 0.05 0.02* 20.11 0.05 0.02* 20.11 0.05 0.02* 0.03 ,0.01* 0.13 0.03 ,0.01* 0.04 0.03 0.23 0.04 0.03 0.23 0.03 0.03 0.24 0.58 ,0.01* 2.95 0.58 ,0.01* 1.69 0.59 ,0.01* 1.68 0.59 ,0.01* 1.69 0.59 ,0.01* 0.22 ,0.01* 21.59 0.22 ,0.01* 20.46 0.22 0.04* 20.47 0.22 0.04* 20.45 0.22 0.04* 0.68 0.27 0.81 0.68 0.23 21.42 0.69 0.04* 21.42 0.69 0.04* 21.43 0.69 0.04*

1 Estimated regression coefficients and P values are shown. *Significant associations, P , 0.05. AA, arachidonic acid; BSID-II, Bayley Scales of Infant Development II; MDI, Mental Developmental Index; MeHg, methyl mercury; PDI, Psychomotor Developmental Index; SES, socioeconomic status.

which biological measures of PUFA status were used. Biological measures of PUFA status could be described as more robust than dietary data and differ in that they reflect various physiologic factors that are known to influence PUFA status, such as mobilization of maternal adipose tissue stores during pregnancy

of children aged 7–9 y reported that associations between intake of n–3 PUFA and performance on some outcomes of the Cambridge Neuropsychological Test Assessment Battery varied by n–6/n–3 ratio (29). Data from these studies, using dietary estimates of PUFA (28, 29), agree with results of our study, in

TABLE 3 Main effects models for prenatal MeHg exposure and PUFA status variables and the MacArthur-Bates CDI and IBQ-R at 20 mo of age1 MacArthur-Bates CDI

Sqrt (vocabulary produced) (n = 1265) b DHA + AA MeHg DHA AA Maternal age Hollingshead SES Child sex (girl) Child age Family status n–6/n–3 ratio MeHg n–6/n–3 ratio Maternal age Hollingshead SES Child sex (girl) Child age Family status

SE

P

Sqrt (vocabulary understood) (n = 1265) b

SE

P

IBQ-R

Total gestures (n = 1265) b

SE

P

Surgency (n = 1264) b

SE

P

Negative affect (n = 1264)

Effortful control (n = 1264)

b

b

SE

P

SE P value

20.03 20.53 2.38 20.07 0.02 1.49 0.76 0.54

0.03 0.37 20.01 0.03 0.82 20.07 0.07 0.27 20.00 0.01 0.93 20.01 0.01 1.76 0.76 1.78 1.31 0.17 10.26 3.37 ,0.01* 0.15 0.31 0.63 20.03 0.42 1.81 0.19 2.07 1.35 0.13 0.65 3.48 0.85 0.46 0.32 0.16 0.17 0.44 0.02 ,0.01* 20.04 0.01 ,0.01* 20.06 0.04 0.10 20.00 0.00 0.33 20.01 0.00 0.01 0.09 20.02 0.01 0.07 0.1 0.02 ,0.01* 0.01 0.00 ,0.01* 0.01 0.00 0.23 ,0.01* 0.70 0.17 ,0.01* 3.22 0.45 ,0.01* 20.04 0.04 0.37 0.04 0.06 0.09 ,0.01* 0.36 0.07 ,0.01* 0.99 0.17 ,0.01* 0.01 0.02 0.38 20.04 0.02 0.27 0.05 0.23 0.2 0.26 0.49 0.53 0.35 0.05 0.05 0.28 20.05 0.07

0.35 20.00 0.01 0.99 0.94 0.15 0.31 0.62 0.70 0.07 0.32 0.82 0.06 0.01 0.00 0.07 0.01* 0.01 0.00 ,0.01* 0.46 0.08 0.04 0.06 0.08 0.01 0.02 0.54 0.46 0.07 0.05 0.14

20.04 20.17 20.07 0.02 1.48 0.76 0.56

0.03 0.28 20.01 0.03 0.82 20.06 0.07 0.33 0.00 0.01 1.00 20.01 0.01 0.07 0.02* 20.16 0.05 ,0.01* 20.54 0.14 ,0.01* 0.00 0.01 0.93 0.01 0.02 0.02 ,0.01* 20.03 0.01 0.02* 20.04 0.04 0.24 20.00 0.00 0.50 20.01 0.00 0.01 0.05* 20.01 0.01 0.10 0.1 0.02 ,0.01* 0.01 0.00 ,0.01* 0.01 0.00 0.23 ,0.01* 0.70 0.17 ,0.01* 3.23 0.45 ,0.01* 20.04 0.04 0.38 0.04 0.06 0.09 ,0.01* 0.39 0.07 ,0.01* 1.05 0.17 ,0.01* 0.02 0.02 0.15 20.04 0.02 0.27 0.04* 0.24 0.2 0.24 0.52 0.53 0.32 0.05 0.05 0.30 20.05 0.07

0.36 20.00 0.01 0.99 0.73 20.01 0.01 0.31 0.06 0.01 0.00 0.05 0.01* 0.01 0.00 ,0.01* 0.46 0.08 0.04 0.06 0.09 0.01 0.02 0.50 0.45 0.07 0.05 0.13

1 Estimated regression coefficients and P values are shown. *Significant associations, P , 0.05. AA, arachidonic acid; CDI, Communicative Development Inventories; IBQ-R, Infant Behavior Questionnaire–Revised; MeHg, methyl mercury; SES, socioeconomic status; Sqrt, square root transformation.

6 of 8

STRAIN ET AL. TABLE 4 Interaction models for prenatal MeHg exposure against BSID-II MDI and PDI at 20 mo of age with PUFA status1 20 mo MDI (n = 1243)

Interaction with n–3 PUFA n–3 PUFA lowest tertile n–3 PUFA middle tertile n–3 PUFA highest tertile Hg by n–3 interaction P value Interaction with n–6/n–3 ratio n–6/n–3 PUFA lowest tertile n–6/n–3 PUFA middle tertile n–6/n–3 PUFA highest tertile Hg by n–6/n–3 ratio interaction P value

20 mo PDI (n = 1241)

MeHg b

SE

P

MeHg b

SE

P

0.09 20.17 20.11

0.16 0.14 0.14

0.59 0.24 0.43 0.47

20.23 20.11 0.37

0.16 0.14 0.14

0.15 0.42 ,0.01* ,0.01*

20.04 20.06 20.12

0.13 0.15 0.17

0.78 0.70 0.48 0.93

0.19 0.14 20.37

0.13 0.15 0.17

0.15 0.36 0.03* 0.02*

Estimated regression coefficients and P values are shown. *Significant associations, P , 0.05. Models shown are adjusted for maternal age, child age, child sex, Hollingshead socioeconomic status, and family status. High n–3 PUFA, .0.308 mg/mL; medium n–3 PUFA, 0.228–0.308 mg/mL; low n–3 PUFA, ,0.228 mg/mL. Serum concentrations for the tertiles were as follows: High n–6/n–3, .4.496; medium n–6/n–3, 3.523–4.496; low n–6/n–3, ,3.523. BSID-II, Bayley Scales of Infant Development II; Hg, mercury; MDI, Mental Developmental Index; MeHg, methyl mercury; PDI, Psychomotor Developmental Index. 1

(30). There is no currently recommended biological n–6/n–3 ratio, but ratios in the current cohort (4.36; range: 1.56–15.81) are lower than most Western populations as a result of a high dietary intake of n–3 PUFA from fish (24). Nevertheless, our results suggest that if there are adverse developmental associations with the effects of prenatal MeHg exposure at these levels, they may depend on the physiologic balance of n–6 to n–3 PUFA. If true, the mechanisms of this association deserve further study. This study also suggests that PUFA status, independent of MeHg, may be associated with improved communication. Increased maternal DHA concentrations were associated with an improved CDI vocabulary understood score. In contrast, increased

maternal n–6/n–3 ratios were associated with poorer scores on all 3 CDI scores. These results suggest a higher maternal DHA status, and a lower n–6/n–3 ratio may be beneficial for child language and communication skills. These data support our earlier finding in the NC1 cohort at age 5 y in which, on another test of language development, the Preschool Language Scale, scores improved with increasing DHA (11). Other studies of children at a similar age and that used the CDI reported that language skills improved with higher n–3 PUFA (12, 31) and maternal fish intake (12). We did not predict lower test scores on the MDI with increasing maternal DHA, and this finding conflicts with findings from the earlier NC1 cohort in which we found no significant effects of PUFA on the MDI (8, 9, 11). One explanation might be

FIGURE 2. Interactions between maternal hair mercury and tertiles of maternal serum n–3 PUFA (mg/mL) (A) and maternal n–6/n–3 ratio (B) on the BSID-II PDI score, with superimposed mercury slopes and tertile-specific P values from the covariate-adjusted regression (n = 1241). The maternal hair mercury slopes within tertiles of maternal serum n–3 PUFA (A) and within tertiles of n–6/n–3 (B) were significantly different from each other (P , 0.01 and P = 0.02, respectively, for the 2 df tests). Serum concentrations for the tertiles were as follows: high n–3 PUFA, .0.308 mg/mL; medium n–3 PUFA, 0.228– 0.308 mg/mL; low n–3 PUFA, ,0.228 mg/mL. High n–6/n–3, .4.496; medium n–6/n–3, 3.523–4.496; low n–6/n–3, ,3.523. BSID-II, Bayley Scales of Infant Development II.

PRENATAL MeHg, PUFA, AND CHILD DEVELOPMENT

7 of 8

FIGURE 3. Associations between the 20-mo McArthur-Bates CDI total gestures score and maternal hair mercury (slope = 20.07, SE = 0.07, P = 0.27) (A), maternal DHA status (mg/mL serum) (slope = 10.26, SE = 3.37, P , 0.01) (B), and maternal n–6/n–3 ratio (slope = 20.54, SE = 0.14, P , 0.01) (C), with superimposed slopes from the covariate-adjusted regression (n = 1265). CDI, Communicative Development Inventories.

that AA becomes the limiting nutrient for some aspects of fetal development when DHA concentrations are high, such as in high fish-eating populations, whereas DHA is the limiting nutrient in populations consuming low quantities of fish. This hypothesis was proposed following studies in Tanzania, where tribes with differing levels of fish consumption were studied. These studies, which assessed erythrocyte PUFA during pregnancy and postpartum, reported a synergistic relation between DHA and AA with low DHA status and an antagonistic relation with high DHA status (32–34). Our study has a number of strengths. The cohort was large and enrolled specifically to address the complex relationships of fish consumption, MeHg exposure, and nutritional status. Mothers were early in pregnancy and followed prospectively. Mothers consumed large quantities of fish, resulting in mean MeHg exposures about 10 times higher than in US (35) or UK women (36). Robust physiologic measures of both MeHg exposure and PUFA status were used as previously described, and the cohort size permitted investigation of interactions between maternal MeHg exposure and PUFA status. The study also has limitations. It was an observational study and thus cannot determine causation, as unmeasured covariates might have confounded some relationships. Missing covariates reduced the size of the cohort available for analysis, although there was no meaningful difference in demographic and exposure characteristics between those who were and were not included. We found no adverse associations between prenatal MeHg exposure from fish consumption and child development through 20 mo of age in this study, even though the mothers consumed large quantities of fish and had exposures about 10 times those of Western nations. These results confirm our past MeHg findings in multiple cohorts from this population, support the importance of maternal PUFA status, and suggest that the developmental consequences of exposure to MeHg from fish consumption and maternal PUFA status are more complex than previously thought. We thank Jean Reeves and Joanne Janciuras from the University of Rochester for their assistance with database management. The authors’ responsibilities were as follows—JJS: had full access to all data in the study, with the exception of mercury data, and took responsibility for data integrity and the accuracy of data analysis, as well as the decision to

submit for publication; JJS, EvW, EMM, GEW, THS, CFS, GJM, and PWD: were involved in study concept, design, and funding acquisition; CFS, JH, KY, and EvW: were involved in fieldwork and acquisition of the data; EvW, SWT, TML, and DH: conducted the statistical analysis and interpretation of data; EvW, AJY, SWT, MSM, GEW, and DH: conducted quality control assessment and were responsible for analysis and interpretation of data; and JJS and AJY: drafted the manuscript; EvW, SWT, MSM, EMM, GEW, TML, THS, KY, DH, CFS, JH, GJM, and PWD: contributed to critical revision of the manuscript. All authors read and approved the final manuscript. All authors declared no conflicts of interest related to this study.

REFERENCES 1. Cohen JT, Bellinger DC, Connor WE, Shaywitz BA. A quantitative analysis of prenatal intake of n–3 polyunsaturated fatty acids and cognitive development. Am J Prev Med 2005;29:366–74. 2. Mozaffarian D, Rimm EB. Fish intake, contaminants, and human health: evaluating the risks and the benefits. JAMA 2006;296:1885–99. 3. Leino O, Karjalainen AK, Tuomisto JT. Effects of docosahexaenoic acid and methylmercury on child’s brain development due to consumption of fish by Finnish mother during pregnancy: a probabilistic modeling approach. Food Chem Toxicol 2013;54:50–8. 4. Zeilmaker MJ, Hoekstra J, van Eijkeren JC, de Jong N, Hart A, Kennedy M, Owen H, Gunnlaugsdottir H. Fish consumption during child bearing age: a quantitative risk-benefit analysis on neurodevelopment. Food Chem Toxicol 2013;54:30–4. 5. Davidson PW, Myers GJ, Cox C, Axtell C, Shamlaye C, Sloane-Reeves J, Cernichiari E, Needham L, Choi A, Wang Y, et al. Effects of prenatal and postnatal methylmercury exposure from fish consumption on neurodevelopment: outcomes at 66 months of age in the Seychelles Child Development Study. JAMA 1998;280:701–7. 6. Myers GJ, Davidson PW, Cox C, Shamlaye CF, Palumbo D, Cernichiari E, Sloane-Reeves J, Wilding GE, Kost J, Huang LS, et al. Prenatal methylmercury exposure from ocean fish consumption in the Seychelles Child Development Study. Lancet 2003;361:1686–92. 7. Davidson PW, Strain JJ, Myers GJ, Thurston SW, Bonham MP, Shamlaye CF, Stokes-Riner A, Wallace JM, Robson PJ, Duffy EM, et al. Neurodevelopmental effects of maternal nutritional status and exposure to methylmercury from eating fish during pregnancy. Neurotoxicology 2008;29:767–75. 8. Strain JJ, Davidson PW, Bonham MP, Duffy EM, Stokes-Riner A, Thurston SW, Wallace JM, Robson PJ, Shamlaye CF, Georger LA, et al. Associations of maternal long-chain polyunsaturated fatty acids, methyl mercury, and infant development in the Seychelles Child Development Nutrition Study. Neurotoxicology 2008;29:776–82. 9. Stokes-Riner A, Thurston SW, Myers GJ, Duffy EM, Wallace J, Bonham M, Robson P, Shamlaye CF, Strain JJ, Watson G, et al. A longitudinal analysis of prenatal exposure to methylmercury and fatty acids in the Seychelles. Neurotoxicol Teratol 2011;33:325–8.

8 of 8

STRAIN ET AL.

10. Davidson PW, Cory-Slechta DA, Thurston SW, Huang LS, Shamlaye CF, Gunzler D, Watson G, van Wijngaarden E, Zareba G, Klein JD, et al. Fish consumption and prenatal methylmercury exposure: cognitive and behavioral outcomes in the main cohort at 17 years from the Seychelles child development study. Neurotoxicology 2011;32:711–7. 11. Strain JJ, Davidson PW, Thurston SW, Harrington D, Mulhern MS, McAfee AJ, van Wijngaarden E, Shamlaye CF, Henderson J, Watson GE, et al. Maternal PUFA status but not prenatal methylmercury exposure is associated with children’s language functions at age five years in the Seychelles. J Nutr 2012;142:1943–9. 12. Daniels JL, Longnecker MP, Rowland AS, Golding J.; ALSPAC Study Team, University of Bristol Institute of Child Health. Fish intake during pregnancy and early cognitive development of offspring. Epidemiology 2004;15:394–402. 13. Llop S, Guxens M, Murcia M, Lertxundi A, Ramon R, Riano I, Rebagliato M, Ibarluzea J, Tardon A, Sunyer J, et al. Prenatal exposure to mercury and infant neurodevelopment in a multicenter cohort in Spain: study of potential modifiers. Am J Epidemiol 2012;175:451–65. 14. Crump KS, Kjellstrom T, Shipp AM, Silvers A, Stewart A. Influence of prenatal mercury exposure upon scholastic and psychological test performance: benchmark analysis of a New Zealand cohort. Risk Anal 1998;18:701–13. 15. Grandjean P, Weihe P, White RF, Debes F, Araki S, Yokoyama K, Murata K, Sorensen N, Dahl R, Jorgensen PJ. Cognitive deficit in 7year-old children with prenatal exposure to methylmercury. Neurotoxicol Teratol 1997;19:417–28. 16. Jedrychowski W, Jankowski J, Flak E, Skarupa A, Mroz E, SochackaTatara E, Lisowska-Miszczyk I, Szpanowska-Wohn A, Rauh V, Skolicki Z, et al. Effects of prenatal exposure to mercury on cognitive and psychomotor function in one-year-old infants: epidemiologic cohort study in Poland. Ann Epidemiol 2006;16:439–47. 17. Sagiv SK, Thurston SW, Bellinger DC, Amarasiriwardena C, Korrick SA. Prenatal exposure to mercury and fish consumption during pregnancy and attention-deficit/hyperactivity disorder–related behavior in children. Arch Pediatr Adolesc Med 2012;166:1123–31. 18. Lynch ML, Huang LS, Cox C, Strain JJ, Myers GJ, Bonham MP, Shamlaye CF, Stokes-Riner A, Wallace JM, Duffy EM, et al. Varying coefficient function models to explore interactions between maternal nutritional status and prenatal methylmercury toxicity in the Seychelles Child Development Nutrition Study. Environ Res 2011;111:75–80. 19. do Nascimento JL, Oliveira KR, Crespo-Lopez ME, Macchi BM, Maues LA, Pinheiro Mda C, Silveira LC, Herculano AM. Methylmercury neurotoxicity & antioxidant defenses. Indian J Med Res 2008; 128:373–82. 20. Janssen CI, Kiliaan AJ. Long-chain polyunsaturated fatty acids (LCPUFA) from genesis to senescence: the influence of LCPUFA on neural development, aging, and neurodegeneration. Prog Lipid Res 2014;53:1–17. 21. Folch J, Lees M, Sloane Stanley GH. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 1957; 226:497–509. 22. Herrera E. Implications of dietary fatty acids during pregnancy on placental, fetal and postnatal development—a review. Placenta 2002;23 (Suppl A):S9–S19.

23. Kjellmer L, Hedvall A, Fernell E, Gillberg C, Norrelgen F. Language and communication skills in preschool children with autism spectrum disorders: contribution of cognition, severity of autism symptoms, and adaptive functioning to the variability. Res Dev Disabil 2012;33:172–80. 24. Calder PC. Polyunsaturated fatty acids, inflammation, and immunity. Lipids 2001;36:1007–24. 25. Weisberg S. Drawing conclusions. In: Balding DJ, Cressie NAC, Fisher NI, Johnstone IM, Kadane JB, Molenberghs G, Ryan LM, Scott DW, Smith AFM, Teugels JL, editors. Applied linear regression, 3rd edition. New York: John Wiley; 2005. p. 69–95. 26. van Wijngaarden E, Thurston SW, Myers GJ, Strain JJ, Weiss B, Zarcone T, Watson GE, Zareba G, McSorley EM, Mulhern MS, et al. Prenatal methyl mercury exposure in relation to neurodevelopment and behavior at 19 years of age in the Seychelles Child Development Study. Neurotoxicol Teratol 2013;39:19–25. 27. Loomans EM, Van den Bergh BR, Schelling M, Vrijkotte TG, van Eijsden M. Maternal long-chain polyunsaturated fatty acid status during early pregnancy and children’s risk of problem behavior at age 5-6 years. J Pediatr 2014;164:762–8. 28. Bernard JY, De Agostini M, Forhan A, de Lauzon-Guillain B, Charles MA, Heude B. The dietary n6:n3 fatty acid ratio during pregnancy is inversely associated with child neurodevelopment in the EDEN mother-child cohort. J Nutr 2013;143:1481–8. 29. Sheppard KW, Cheatham CL. Omega-6 to omega-3 fatty acid ratio and higher-order cognitive functions in 7- to 9-y-olds: a cross-sectional study. Am J Clin Nutr 2013;98:659–67. 30. Haggarty P. Fatty acid supply to the human fetus. Annu Rev Nutr 2010; 30:237–55. 31. Meldrum SJ, D’Vaz N, Simmer K, Dunstan JA, Hird K, Prescott SL. Effects of high-dose fish oil supplementation during early infancy on neurodevelopment and language: a randomised controlled trial. Br J Nutr 2012;108:1443–54. 32. Kuipers RS, Luxwolda MF, Janneke Dijck-Brouwer DA, Muskiet FA. Intrauterine, postpartum and adult relationships between arachidonic acid (AA) and docosahexaenoic acid (DHA). Prostaglandins Leukot Essent Fatty Acids 2011;85:245–52. 33. Luxwolda MF, Kuipers RS, Sango WS, Kwesigabo G, Dijck-Brouwer DA, Muskiet FA. A maternal erythrocyte DHA content of approximately 6 g% is the DHA status at which intrauterine DHA biomagnifications turns into bioattenuation and postnatal infant DHA equilibrium is reached. Eur J Nutr 2012;51:665–75. 34. Luxwolda MF, Kuipers RS, Smit EN, Velzing-Aarts FV, DijckBrouwer DA, Muskiet FA. The relation between the omega-3 index and arachidonic acid is bell shaped: synergistic at low EPA+DHA status and antagonistic at high EPA+DHA status. Prostaglandins Leukot Essent Fatty Acids 2011;85:171–8. 35. McDowell MA, Dillon CF, Osterloh J, Bolger PM, Pellizzari E, Fernando R, Montes de Oca R, Schober SE, Sinks T, Jones RL, et al. Hair mercury levels in U.S. children and women of childbearing age: reference range data from NHANES 1999-2000. Environ Health Perspect 2004;112:1165–71. 36. Lindow SW, Knight R, Batty J, Haswell SJ. Maternal and neonatal hair mercury concentrations: the effect of dental amalgam. BJOG 2003; 110:287–91.