Study Areas and Subjects

Da Nang city is located in the middle of Vietnam and Da Nang airbase is a known hot spot of dioxin contamination. The Thanh Khe and Son Tra districts in Da Nang City, located within 10 km of Da Nang airbase, were selected as the study locations. In collaboration with the Vietnamese government, Hatfield Consultants monitored environmental hazards, including heavy metals and organochlorines in the soil of the area surrounding Da Nang airbase, reporting that dioxins are prominent among the hazardous chemicals found there [2].

Enrolled subjects were children whose mothers were living in the Thanh Khe and Son Tra districts in Da Nang city. They were recruited at birth by obstetricians at district hospitals in Thanh Khe and Son Tra. The criteria for recruitment of mothers at the hospitals were as follows: i) mothers must have resided in one of the study districts for a period encompassing at least the duration of their pregnancy; ii) mothers must have given birth to full-term babies; and iii) there must have been no complications during childbirth. A total of 241 mother–infant pairs were enrolled at baseline, including 159 mothers that gave birth in the Thanh Khe district hospital in 2008 and 82 mothers that gave birth in the Son Tra district hospital in 2009. Subsequently, follow-up examinations were conducted at 4 months, 1 year, and 3 years after birth. Of 241 mother–infant pairs at baseline, 78.0% (188 pairs; 120 in Thanh Khe and 68 in Son Tra) participated in the 5-year follow-up assessment, conducted in Thanh Khe in December 2013 and in Son Tra in March 2015. Fifty-three pairs did not receive a follow-up assessment because they moved to other areas or were absent on test days. Of children who participated in the follow-up, mother’s information was missing for seven children. Thus, the final sample for the analysis included 181 children (115 and 66 children from Thanh Khe and Son Tra districts, respectively), which represented 75.1% of the original sample. No significant differences were found regarding characteristics or breast milk dioxin levels between participants and drop-outs, or included and excluded participants. Written informed consent to participate in the survey and to publish data was obtained from all mothers according to a process reviewed and approved by the Health Department of Da Nang City. The study design was approved by the Institutional Ethics Board for Epidemiological Studies at Kanazawa Medical University (No. 187).

Interview and Growth Parameters

Information was collected from mothers (age, residential history, parity, smoking habits of the mother and other family members, alcohol consumption, education, and family income) and the children (age at the assessment, gestational weeks, and gender). Physical parameters of the children, including weight, height, and head and abdominal circumferences, were measured at birth and 5 years, and mean (SD) of measurement values are shown in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of subjects in three groups with different exposure levels of TEQ-PCDDs/Fs in breast milk.

https://doi.org/10.1371/journal.pone.0147655.t001

Breast Milk Collection and Measurement of PCDDs/Fs in Breast Milk

A breast milk sample was collected from each nursing mother 1 month after birth by hand expression with the assistance of a midwife or medical worker. Samples were placed in clean polyethylene containers and temporarily stored at − 4°C in local health centers. Samples were frozen on dry ice for transport to Japan by airplane, and were analyzed at the Center for High Technology, Kanazawa Medical University (Uchinada, Japan). Approximately 10 mL of breast milk was used to quantify levels of 17 2,3,7,8-substituted PCDDs/Fs, including seven congeners of PCDDs and 10 congeners of PCDFs, using a gas chromatograph (HP-6980; Hewlett-Packard, Palo Alto, CA, USA) equipped with a high-resolution mass spectrometer (MStation-JMS700, JEOL, Tokyo, Japan). The methods and results of dioxin analysis have been described previously [3, 8]. Briefly, 17 congeners of PCDDs/Fs were quantified and concentrations in picograms per gram of lipid (pg/g lipid) were calculated. TEQ of PCDDs/Fs was calculated as the sum of all values obtained, multiplying each congener concentration by its TEQ factor from the WHO 2005-TEF [14]. Values of congeners with concentrations below detection limits were set to half the detection limit.

Subjects were divided into three groups (high, moderate, and low exposure groups) according to TCDD and TEQ-PCDDs/Fs levels in breast milk as follows: a) TCDD low <2.5, medium 2.5–3.5, and high ≥3.5 pg/g lipid; and b) TEQ-PCDDs/Fs low <11.5, medium 11.5–17.6, and high ≥17.6 pg-TEQ/g lipid. For analysis of KABC-II scores in children aged 60 months or older, the subjects were divided into two groups according to the TCDD in breast milk with a cutoff level of 2.5 pg/g lipid.

Neurodevelopmental Assessments

In the present study, we used two neurodevelopment assessment batteries for children, Movement ABC-2 and KABC-II. Children were tested in a community health station by one examiner for each test in the presence of the child’s caregiver. The examiners were blinded to exposure levels and previous neurodevelopmental outcomes of each subject. During the Movement ABC-2 testing, 5 children refused to cooperate and the final sample for analysis of coordination movement included 176 children, although all children were participated KABC-II.

Movement ABC-2 (Pearson Education Inc., UK) is a measure of manual dexterity (MD), aiming and catching (A&C), and balance (Bal) of children over 3 years old. The MD scale assesses skills such as posting coins, threading beads, and drawing trails. The A&C scale assesses skills such as catching beanbags and throwing beanbags onto mats. The Bal scale assesses skills such as one-leg balance, walking with heels raised, and jumping on mats. For a total MD score, we added the three subscale scores, for a total A&C score, we added the two subscale scores. For the total Bal score, we added the three Bal subscale scores. Finally, we added each total score to obtain a total test score (TOTAL).

Cognitive abilities of the children were assessed using KABC-II (NCS Pearson Inc., USA). Cognitive functions used in the present study included conceptual thinking, face recognition, number recall, triangles (shape matching), word order, pattern reasoning, and hand movements. Nonverbal index (NVI), sequential/global scales of short-term memory (Seq/Gsm), and simultaneous/global scales of visual processing (Sim/Gv) were used as global scales. Scores for conceptual thinking, face recognition, triangles, pattern reasoning, and hand movements for 5-year-olds, and scores for four of these subscales excluding pattern reasoning for 4-year-olds were combined to provide a NVI score. Number recall and word order scores were summed to provide a score of Seq/Gsm. Finally, Sim/Gv score was the total score of conceptual thinking, triangles, and pattern reasoning tests for 5-year-olds, with the latter replaced by face recognition for 4-year-olds.

Data Analysis

SPSS software (Version 21.0,; IBM Corp., Armonk, NY, USA) was used for statistical analyses. The concentrations of dioxins in breast milk were base-10 logarithmically transformed to improve normality. A general linear model was used to compare Movement ABC-2 and KABC-II scores of the three TEQ-PCDDs/Fs or TCDD exposure groups after adjusting for the following covariates: parity (primipara or multipara), maternal age, family members’ smoking habits, family income (Vietnamese Dong), maternal alcohol consumption during pregnancy (yes or no), maternal education (years), gestational age at birth, birth weight, age at present examination (months), and area of residence (Thanh Khe or Son Tra). Single variable tests were used to test significant difference among exposure groups. Results were considered significant if P <0.05.

Comparisons of Subject Characteristics among TEQ-PCDDs/Fs Exposure Groups

Among the 176 mother-children pairs participated in neurodevelopmental assessments at 5 years old, we compared the mean and standard deviation (SD) of characteristics between the three TEQ-PCDDs/Fs exposure groups. The results split by gender are shown in Table 1. In both boys and girls, there was no significant difference in the characteristics of mothers or children among the three TEQ-PCDDs/Fs exposure groups, except for maternal age, which was significantly higher in the high TEQ-PCDDs/Fs group compared with the low exposure group. Levels of TCDD were significantly increased with increasing levels of TEQ-PCDDs/Fs in these three groups (Table 1).

Comparisons of Movement ABC-2 Scores among TEQ-PCDDs/Fs and TCDD Exposure Groups

Table 2 shows the adjusted Movement ABC-2 scores in the three TEQ-PCDDs/Fs exposure groups for each gender; high (≥17.6 pg-TEQ/g lipid), moderate (11.5–17.5 pg-TEQ) and low (<11.5) exposure groups. In boys, the TOTAL score was significantly lower in the high TEQ-PCDDs/Fs exposure group than in the moderate and low exposure groups. For subscales, total BAL score was significantly lower in the high TEQ-PCDDs/Fs group compared with the moderate and low exposure groups in boys. Among the three BAL scores, scores for jumping on mats contributed to the decreased total BAL score for boys in the high TEQ-PCDDs/Fs exposure group. The score for walking with heels raised was also lower in the high TEQ-PCDDs/Fs exposure group, but this difference in scores was exhibited only between the moderate and high TEQ-PCDDs/Fs exposure groups in boys. In girls, no significant differences in Movement ABC-2 scores were found among the three TEQ-PCDDs/Fs exposure groups. Additionally, there was no significant difference in Movement ABC-2 scores among the three TCDD exposure groups in children of any gender (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Comparisons of adjusted Movement ABC-2 scores among the three TEQ-PCDDs/Fs exposure groups.

https://doi.org/10.1371/journal.pone.0147655.t002

Comparisons of KABC-II Scores among TEQ-PCDDs/Fs and TCDD Exposure Groups

Table 3 shows comparisons of adjusted KABC-II scores among the high (≥ 3.5 pg/g lipid), moderate (2.5–3.5 pg/g lipid), and low (< 2.5 pg/g lipid) TCDD groups. In boys, NVI scores were significantly lower in the high TCDD exposure group compared with the low exposure group, although there was no significant difference in NVI scores in girls. There were no significant differences in Seq/Gsm and Sim/Gv scores among the three TCDD exposure groups in either gender. Subsequently, NVI scores were compared between the three TCDD exposure groups in children of 5 years or older, as NVI score is assessed differently in 5-year-olds, including the pattern reasoning assessment. In boys, NVI scores in 5-year-olds were significantly lower in the high TCDD exposure group compared with the low exposure group (Table 3). However, there were no significant differences between TCDD exposure groups in Seq/Gsm and Sim/Gv scores in boys or in any scores in girls. Additionally, no significant differences in cognitive global scores were found between the three TEQ-PCDDs/Fs exposure groups in all children or the subgroup of 5-year-olds of any gender (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Comparisons of adjusted KABC-II scores among the three TCDD exposure groups.

https://doi.org/10.1371/journal.pone.0147655.t003

To compare KABC-II subscale scores, 5-year-olds were divided into two groups: low (< 2.5 pg/g lipid) and high (≥ 2.5 pg/g lipid). The cutoff of 2.5 pg/g lipid was used, because the number of boys with TCDD ≥ 3.5 pg/g lipid was small. In boys, NVI scores were significantly lower in the high TCDD exposure group compared with the low exposure group (Table 4). Boys in the high TCDD group also had lower Sim/Gv scores, but this difference was not significant (P = 0.056). Among subscales of KABC-II scores, the pattern reasoning contributed to the decreased NVI score in the high TCDD exposure group (Table 4). In boys, face recognition scores were lower in the high TCDD exposure group, but this difference was not significant. These results suggest that decreased pattern reasoning scores were the biggest contributor to decreased NVI scores in boys.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Comparisons of adjusted KABC-II scores between high and low TCDD exposure groups in 5-year-olds.

https://doi.org/10.1371/journal.pone.0147655.t004

Comparison of the Effects of TEQ-PCDDs/Fs and TCDD on Neurodevelopment

To investigate the relationship between dioxin exposure and neurodevelopmental impairments, children were divided into three risk groups for impairment according to the number of risks present for difficulties in motor coordination and/or cognitive function (no risk, single risk of either function, or double risk of both functions). We compared the rate of children with TEQ-PCDDs/Fs ≥17.6 pg-TEQ/g lipid and TCDD ≥2.5 pg/g lipid in the three risk groups for neurodevelopmental impairment (Table 5). We used the guidebook for each assessment to define cutoff values for being at risk of neurodevelopmental impairment. For the Movement ABC-2, this was the 15^th percentile of the general population (67 points) [15]. For the NVI scale, this was -2SD of the general population (70 points) [16]. Five boys (5%) and one girl had double risks for difficulties in both neurodevelopmental skills, while 25 boys (25%) and 12 girls (15.8%) had a single risk for impairment in either motor coordination or cognitive skill. In boys, children at double risk had 100% higher exposure rates for TEQ-PCDDs/Fs exposure and 80% for TCDD exposure. These rates were significantly higher compared with children at no risk for impairment (Table 5). In boys at single risk, only the rate of TEQ-PCDDs/Fs ≥17.6 pg-TEQ/g lipid was significantly higher compared with children at no risk (Table 5). Additionally, 23 boys (92%) at single risk had difficulties in motor coordination. However, none of these effects were found for girls.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Children at risk for having difficulties in motor coordination and/or cognitive ability.

https://doi.org/10.1371/journal.pone.0147655.t005

Effects of Dioxin on Motor Coordination and Higher Cognitive Abilities

In the present study, scores for motor coordination, particularly balance, were significantly lower in preschool boys exposed to high levels of TEQ-PCDDs/Fs during the perinatal period, although movement impairment was not associated with TCDD exposure. In contrast, NVI scores (indicating cognitive ability) were lower in boys exposed to high levels of TCDD, but not TEQ-PCDDs/Fs. In the present subjects, the most contributed congeners to TEQ-PCDDs/Fs in breast milk were 1,2,3,7,8-PentaCDD (34%), 2,3,4,7,8-PentaCDF (18%), and 1,2,3,6,7,8-HexaCDF (14%), while the contribution rate of TCDD was 12% of TEQ-PCDDs/Fs. These congeners, particularly 2,3,4,7,8-PentaCDF, have different toxic effects compared with TCDD in human tissue or cells [17, 18]. Thus, the effects of TEQ-PCDDs/Fs on motor coordination might be caused by other PCDDs/Fs congeners than TCDD. However, boys at double risk for both movement and cognitive difficulties tended to be those with higher exposure to both TEQ-PCDDs/Fs and TCDD, suggesting that high levels of total dioxin exposure with increased TCDD may increase the risk of cognitive impairment combined with motor coordination problems similar to children with ASD [12,13].

Impact of Dioxin Exposure on Motor Development in Previous Studies

In previous assessments of the present cohort at 3 years old, we reported that motor scores (assessed using the Bayley III instrument), cognitive measures, and language scores were all lower in boys exposed to high TEQ-PCDDs/Fs (≥17.6 pg-TEQ/g lipid) [11]. It is possible that boys in the high TEQ-PCDDs/Fs exposure group who showed poor general motor skills at 3 years old, also exhibited poor motor coordination skills at five years old, suggesting that TEQ-PCDDs/Fs exposure increases the risk for developmental coordination disorder (DCD). Indeed, children with DCD often show delays in achieving motor milestones when they are younger [19]. The Movement ABC is a widely used test of motor skills for school-age children, and is also used to detect motor outcomes of children exposed to environmental pollutants. Roze et al. (2009) administered the Movement ABC, version 1, to 62 children in the Netherlands to investigate the influence of prenatal exposure to organohalogen compounds, such as hydroxylated polychlorinated biphenyls (OH-PCBs), on motor development. They reported an inverse correlation between OH-PCBs and fine manipulation abilities (i.e. a low MD score) [20]. However, balance control, which was particularly affected by TEQ-PCDDs/Fs exposure in the present study, is a sensorimotor integration skill often impaired in children with ASD [21–24]. These motor effects may arise from a common mechanism in the brain, as high TCDD exposure has previously been shown to increase autistic traits at 3 years of age [11]. However, some boys at single risk for movement difficulties were exposed to high levels of TEQ-PCDDs/Fs but not high TCDD. High TEQ-PCDDs/Fs exposure without increased TCDD could occur in any areas exposed to dioxins from causes other than Agent Orange. These results suggest that such exposure might increase the risk of motor coordination difficulties specifically. This might be caused by alteration of reaction time of visual processing due to impairment of myelination in the central nervous system [25]. Our results provide a new perspective on adverse health effects of perinatal dioxin exposure in preschool children, as increased risk for DCD resulting from dioxin exposure has not previously been reported.

Impact of Dioxin Exposure on Higher Cognitive Ability in Children: Present and Previous Studies

To examine cognitive abilities of children, we used the NVI of KABC-II to derive a total global scale score, because only this scale was validated for a non-English-speaking population. Among the five subscales of the NVI for 5-year-olds, pattern reasoning scores contributed most to decreased NVI scores in boys exposed to high TCDD (≥2.5 pg/g lipid). This suggests that poor planning ability is related to increased TCDD exposure. Scores for face recognition were also lower in boys in the high TCDD exposure group, although the difference in scores was not significant compared with low TCDD groups. These findings suggest that poor cognitive ability observed in the present subjects might be similar to that in children with ASD, who have difficulties in planning and face recognition [26]. This speculation is also supported by the result that TCDD exposure did not affect Seq/Gsm scores, including number recall and word order, both of which are poor in children with mental retardation. In our previous study of the present subjects at 3 years old, we reported increased autistic traits in the high TCDD exposure group (TCDD ≥ 3.5 pg/g lipid) in both genders [11]. However, in the present study conducted at 5 years of age, boys in the higher TCDD group (with a lower cutoff value of 2.5 pg/g lipid) showed increased risk for ASD-like cognitive deficits. Since testing for higher cognitive abilities is more appropriate at 5 than at 3 years old, we suggest that we have detected mild effects of dioxin exposure on cognitive ability in boys exposed to moderately increased TCDD. Future investigations are required to investigate whether the present cohort shows deficits similar to those in children with ASD in higher cognitive abilities, such as executive functions, including planning and theory of mind [27].

A Dutch cohort study of school-age children reported an inverse relationship between dioxin exposure and both visual motion processing and cognition [28], but no relationship between dioxin exposure and neurodevelopmental test results, including IQ [25]. This suggests that effects of dioxin exposure on cognitive functions in children might only be evident in specific higher brain functions. Consequently, they can only be detected by specific psychological tests and neurophysiological examinations, but not by general developmental tests. Therefore, we are planning to follow-up these children and assess alterations in specific higher cognitive functions, using neurophysiological tests such as EEG responses to visual and auditory stimuli.

Effects of Dioxins on Higher Brain Function and Mechanisms in Animal Studies

In animals, we reported that prenatal TCDD exposure induces socio-emotional deficits particularly in male offspring owing to alterations in Ca²⁺/calmodulin-dependent protein kinase IIα (CaMKIIα) activity and decreased number of parvalbumine-positive neurons (putative inhibitory interneurons) in the limbic system of rat brains [29,30]. These results suggested that perinatal TCDD exposure influences social brain function, a higher brain function, that is also observed in children with ASD. Moreover, Endo et al. (2012) reported behavioral inflexibility, compulsive repetitive behavior, and lower competitive dominance in mice exposed to TCDD during the perinatal period using IntelliCage apparatus and suggested that perinatal TCDD exposure may lead to develop executive function deficits and social behavioral abnormality [31]. In rat offspring exposed to TCDD during the perinatal period, Kakeyama et al. (2014) also demonstrated poor paired-associate learning behavior, suggesting that perinatal TCDD exposure disrupted higher brain function [32]. These higher brain function deficits were indicated to be accompanied with alteration of neuronal activity in the medial prefrontal cortex and amygdala of the limbic system of mice brains [31]. These reports are consistent with our finding of lower cognitive scores in 5-year-old children with high TCDD exposure in the present study. Further studies are required to determine the exact neuronal alterations in the brain, and effects on neuronal networks, particularly limbic system of the brain.

Gender Difference

Consistent with our previous studies of the birth cohort in Da Nang at 4 months [9] and 3 years of age [11], gender differences in the effects of dioxin on neurodevelopment were evident in the present study. Our results suggest that boys are more susceptible to dioxin toxicity than girls. Higher prevalence of ASD and DCD in boys is also well established [20, 33]. The interaction between genes and the environment leading to early exposure to androgenic hormones is a potential etiological mechanism affecting sex-specific susceptibility to ASD [34]. An inverse association between testosterone levels in cord blood and maternal di (2-ethylhexyl) phthalate exposure, a persistent organic pollutant, was reported in a Japanese birth cohort [35]. These results suggest that maternal dioxin exposure might influence testosterone levels of the fetus which play an important role in autistic traits in childhood.

A study in the Netherlands reported endocrine disrupting effects of perinatal PCDDs/Fs and PCBs exposure on play behaviors exhibited by children, with more feminized play observed in exposed boys at 8 years of age [36]. Recently, Winnike et al. (2014) also investigated sex-typical play behaviors of German children aged 6–8 years and reported that perinatal exposure to PCDDs/Fs and PCBs might modify behavioral sexual dimorphism in children [37]. They suggested that dioxins interact with the hypothalamic-pituitary-gonadal axis in the brain to induce endocrine-disrupter effects. In the same German cohort, it was also reported that prenatal and perinatal exposure to PCDDs/Fs and PCBs influenced attention-deficit hyperactivity disorder—related behavior [38]. These results suggest that male susceptibility to the effects of dioxin on neurodevelopment might be mediated through an interaction of dioxins with the hypothalamic-pituitary-gonadal axis, consequently affecting androgenic hormone levels in the fetal brain. Our future studies following up these children will investigate associations between dioxin exposure and sexual dimorphism, in addition to cognitive abilities.

Limitations

The validity of the results and conclusions of the present study must be considered within the context of the study’s strengths and limitations. Although the Movement ABC-2 and KABC-II are useful tests for making independent quantitative judgments regarding motor coordination and cognitive ability, they have some limitations with respect to the present study. The Movement ABC-2 was developed and standardized for infants in the United Kingdom, not for Vietnamese children. Therefore, we could not judge the developmental levels of individual children based on the scores obtained with this test. However, we used their cutoff value to assess risk for movement difficulties, resulting in possible overestimation of children with possible DCD. Regarding KABC-II, it is possible that we overestimated the number of children at risk for cognitive impairment, albeit we used the NVI, which is validated for non-English-speaking populations. However, each test was performed by a single examiner who was well trained and followed specific instructions as described in the manual. Therefore, the comparison of test scores among the different exposure groups within the same population, to clarify associations between the levels of dioxin exposure and test scores of the test should be reliable.