Poor psychometric scores of children living in isolated riverine and agrarian communities and fish–methylmercury exposure

Poor psychometric scores of children living in isolated riverine and agrariancommunities and fish–methylmercury exposure

Marlon de F. Fonseca a,b, Jose G. Dorea c,*, Wanderley R. Bastos d, Rejane C. Marques a,d,Joao P.M. Torres a, Olaf Malm a

a Laboratorio de Radioisotopos Eduardo Penna Franca, Instituto de Biofısica Carlos Chagas Filho, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ, Brazilb Instituto Fernandes Figueira, FIOCRUZ, Rio de Janeiro, RJ, Brazilc Universidade de Brasılia, Faculty of Health Sciences, C.P. 04322, Campus - Asa Norte, 70919-970 Brasılia, DF, Brazild Laboratorio de Biogeoquımica Ambiental Wolfgang C. Pfeiffer, Universidade Federal de Rondonia, Porto Velho, RO, Brazil

NeuroToxicology 29 (2008) 1008–1015

A R T I C L E I N F O

Article history:

Received 28 January 2008

Accepted 10 July 2008

Available online 25 July 2008

Keywords:

Methylmercury

Fish

Neurocognition

Neurodevelopment

A B S T R A C T

Because of heavy dependence on fish, Amazonian riparian communities are chronically exposed to high

levels of methylmercury (MeHg). We studied fish–MeHg exposure (total hair-Hg, HHg) as a determinant

of neurocognitive scores of children living in two geographically distant, culturally distinct and isolated

poor communities of non-urban environments: Amazonian riverines (Riparians, n = 38) of the Puruzinho

Lake community in the Rio Madeira Basin and rural agrarians from Iuna, Espırito Santo (Agrarians, n = 32).

Nutritional status was estimated by anthropometry (Z-scores) and individual cognitive abilities were

assessed by the Wechsler Intelligence Scale for Children-III (WISC-III) and the Human Figure Drawings

(HFD), both validated versions for Brazilian children. Anthropometric assessment showed slightly

elevated Z-scores for the Agrarian children (not statistically significant) but median HHg concentrations

were 14.4 and 0.25 mg g�1 respectively for Riparian and Agrarian children (p = 0.000). Despite

paradoxical MeHg exposures, both groups showed comparable HFD scores but very poor performance

in WISC-III test battery; median of sum of WISC-III subtests scores (STOT) were 17.9 and 28.6 (p < 0.000)

for Riparian and Agrarian children, respectively (percentage scale). Spearman correlation between

nutritional status (attained growth) and psychometric scores were statistically significant between

height-for-age Z-score and Object Assembly subtest (r = 0.269; p = 0.043), STOT (r = 0.319; p = 0.016),

Performance-IQ (r = 0.311; p = 0.019) and Perceptual Organization Index scores (r = 0.302; p = 0.023). In

these isolated communities there are stronger determinants of neurocognitive poor performance than

MeHg exposure. Global strategies for reducing human exposure to MeHg by curtailing fish consumption

are unrealistic options for riverine subsistence populations and are not justifiable to prevent low

cognitive scores.

� 2008 Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

NeuroToxicology

1. Introduction

Mercury (Hg) is a widely recognized neurotoxic element; lately,warnings have linked environmental Hg exposure and risk ofdiminished cognitive function (Trasande et al., 2005; Cohen et al.,2005; Palmer et al., 2006; Axelrad et al., 2007). Doubling of cord-blood mercury level has been associated with a significantdevelopmental delay for several domains of cognitive function(Grandjean et al., 1997). Studies have suggested that an incrementof 1 ppm in maternal hair (during gestation) can cause a decrease

* Corresponding author. Tel.: +55 61 3368 3575; fax: +55 61 3368 5853.

E-mail address: [email protected] (J.G. Dorea).

0161-813X/$ – see front matter � 2008 Elsevier Inc. All rights reserved.

doi:10.1016/j.neuro.2008.07.001

in IQ to range from 0.18 points (Axelrad et al., 2007) to 0.7 points(Cohen et al., 2005). Therefore there are concerned scientistsworking on advisories for reducing fish–MeHg consumption forAmazonian populations (Passos et al., 2007).

Subsistence populations living along the banks of the Amazo-nian Rivers (ribeirinhos) depend heavily on fish as the dominantfood in their diet (Dorea, 2004). Despite high fish consumption andmethylmercury (MeHg) exposure during pregnancy and through-out life (Dorea et al., 2003; Bastos et al., 2006) there are noestablished endemic neurological diseases associated with fish–MeHg or other neurotoxic components of starchy diets in theseriverine populations (Dorea, 2004).

Because of their heavy dependence on fish, Amazonian riparianwomen and their children (Barbosa et al., 1998) are exposed to

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–1015 1009

high levels of fish–MeHg. Global strategies for reducing humanexposure to MeHg by curtailing fish consumption are unrealdecisions for riverine subsistence populations (Dorea, 2003); thesepopulations consume fish on a daily basis (frequently more thanone fish meal) and depend on it as their main animal-proteinsource. Despite the fact that riverine populations suffer from moredramatic and life-threatening health hazards like endemic malariaand parasitemia, neurodevelopmental outcomes remain an issueof concern (Passos et al., 2007).

We took advantage of our ongoing studies on health issues ofAmazonian riverines to compare neurocognitive performance as afunction of fish consumption. The environmental differences(based on fish–MeHg exposure) between an Amazon riverinecommunity and a rural agrarian community were examined. Whileboth communities are relatively isolated and traditional, riparianvillagers subsist on fish while rural agrarian children rarely everhave fish in their diet.

2. Materials and methods

2.1. Background contextualization

The community of Puruzinho Lake is located in the municipalityof Humaita (Amazonas State) near the border with Rondonia State(Fig. 1). This community of 163 souls inhabits the margin of thelake in 20 households; the housing is rudimentary with no watersupply system, sewage treatment, or electricity; they lead isolatedlives (in places of difficult access) typical of the Amazonianriverines. They are connected to Humaita (32,000 people) only byriver. Intestinal parasites and malaria are endemic and significanthealth problems for this population. Basic education is provided by

Fig. 1. Map of Brazil illustrating distance and geograp

one 4-y-elementary school installed in a one-room building. Twolocal teachers working in one shift each (morning and afternoon)run the 4-h class for all grades. The single classroom is very simple:there are two blackboards on opposite sides of the room that servefor the assignments of the lower and higher grades, while teachingmaterials are absolutely basic (books, glue, paper and colorpencils). Both teachers from Puruzinho Lake completed their ownhigh-school education through programs exclusively developed inisolated Amazonian riverine schools to teach teachers to read andwrite.

The Iuna municipality is located in the Southern region ofEspırito Santo State (Fig. 1) and has a total population of 26,000;half of them are scattered across 460 Km2 of farmland in individualhouseholds or very small communities. The main economicactivities of the region are coffee planting and cattle ranging;the families of the studied children make a living on subsistenceagriculture (corn, beans, etc.) and raising small domestic animals.There are no industries but there are small-scale milk- and meat-processing plants; people usually construct their houses on low-level areas, near clean water sources, which are traditionallyprotected among trees. These rural families can easily travel totown and their social isolation is much less than the Riparianfamilies. Typical coffee crops are characterized by their need forintensive labor, by the use of low technology, and mountainoustopography. Flat lands are preferentially used as cow pasture or forother cultivars. Four rural schools located at various distances fromthe town of Iuna were chosen based on previous contact and staffwillingness to collaborate.

The Iuna selected schools were (1) Escola Municipal de Ensino

Fundamental Deolinda Amorim de Oliveira (located next to the townwith one teacher for each school grade working in independent

hical differences between the two communities.

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–10151010

classrooms; this is the most structured one, and besides havingbasic teaching materials it is equipped with audiovisual resources),(2) Fazenda Boa Vista (located about 2 Km from the town; only oneteacher works the four school grades in the same classroom withtwo opposite blackboards; it has one mimeograph, books, somemaps, simple games, and basic teaching materials), (3) Santo

Antonio das Perobas do Alto Trindade (located more than 10 Kmfrom the town; organization and teaching resources are similar toschool 2), and (4) Serrinha (located more than 10 Km from thetown; like in Puruzinho, two teachers work there in two differentclassrooms: each one uses two opposite blackboards in parallel,one for each of two school grades; organization and teachingresources are similar to schools 2 and 3). Teachers from EspıritoSanto public schools are chosen after completion of regular high-school course with added teaching skills.

Children of both communities (Puruzinho and Iuna) may beexposed to pesticides. Although DDT has been discontinued in theAmazon, pyrethroid spraying is used for vector control of endemicmalaria in Puruzinho; additionally the Rio Madeira receivesagricultural pesticides from its upper basin tributaries. In theIuna communities, environmental exposure to agricultural pesti-cides may also occur as a result of spraying coffee-tree cultivars.Pesticide application is irregular, no more than three times a year,and only during the rainy season, depending on the balance ofprofit and costs for that year. However, women and children do notwork in pesticide-spray activity.

All children (both groups) participating in the study werecurrently enrolled in their respective school; the age range wasvery close (Table 1) with similar mean age (10.2 y). Anothercommonality shared by children of both communities is widerspread in ages (15.5–6.5 y for Riparians and 14.5–7.5 y forAgrarians) compared to the narrower 4 y of elementary schooling.In these remote and impoverished regions, it is common forchildren to remain in school for lack of opportunities of advancingfurther their precarious elementary education. The age distribu-tion of tested children and respective school classes were asfollows; Riparians: 6.5–13.5 y (first grade), 6.5–10.5 y (secondgrade), 8.5–13.5 y (third grade), 8.5–15.5 y (fourth grade);Agrarians: 7.5–9.5 y (first grade), 7.5–11.5 y (second grade), 8.5–14.5 y (third grade), 9.5–14.5 y (fourth grade). It seems that

Table 1Summary of characteristics of children from the riverine Amazonian community of

Puruzinho (Riparian) and the rural Agrarians of Iuna

Groups Riparians (n = 38) Agrarians (n = 32)

Age (years)

Mean (S.D.) 10.2 (2.9) 10.2 (1.5)

Median (Max–Min) 9.5 (15.5 to 6.5) 10.5 (14.5 to 7.5)

School level (years)

Mean (S.D.) 2.6 (1.3) 2.7 (1.0)

Median (Max–Min) 3 (4 to 1) 3 (4 to 1)

Height (Z-score)

Mean (S.D.) �0.834 (1.104) �0.717 (1.007)

Median (Max–Min) �0.69 (1.18 to 3.56) �0.78 (1.34 to 2.87)

Weight (Z-score)

Mean (S.D.) �0.919 (0.936) �0.784 (1.173)

Median (Max–Min) �0.86 (0.60 to �3.11) �0.92 (1.47 to �4.39)

BMI (Z-score)

Mean (S.D.) �0.543 (0.791) �0.497 (1.054)

Median (Max–Min) �0.30 (1.14 to �2.76) �0.43 (1.39 to �3.27)

Hair-Hg* (mg g�1)

Mean (S.D.) 18.4 (12.76) 0.28 (0.22)

Median (Max–Min) 14.4 (57.0 to 3.4) 0.25 (1.23 to 0.05)

*Statistically significant (p = 0.000) with the nonparametric Mann–Whitney U-test

(2-tailed). BMI = body mass index; Weight, Height and BMI Z-scores for age and sex.

Riparian (Puruzinho) children start school earlier. In this remotecommunity, school is the only social and cultural activity available.Attendance and performance are not strictly enforced with somechildren dropping out temporarily or failing exams. This pattern isnot much different in the Agrarian community (Iuna); in thiscommunity school-lunch program seems to make a difference inschool attendance (Fig. 2).

The differences between the Riparians and the Agrarians of Iunacan be noted in many environmental aspects, but the most salientto the present study are MeHg exposure derived from traditionaldominance of fish consumption and pronounced isolation fromurban influences in Puruzinho fishing village. While the Puruzinhosurvival strategy depends exclusively on the aquatic Amazonenvironment, the Iuna community leads a subsistence lifecomparable to many small farmers around the world and doesnot depend on fish. Both groups of children are poor, living inrelatively isolated settings where behavioral consequences offacilitative influences in the environment are restricted. It is fair tosay that Riparian children grow in even more restrictedpsychosocial environment than the Agrarian children.

This study has been approved by a Human Research EthicsCommittee (NESC/UFRJ No. 026/02 and 001/05). After clearexplanation in plain language, the parents of all the childrenprovided written informed consent. A simple questionnaire wasused to assess children’s biodata and general health aspects; theseincluded: parents’ schooling, use of medication, vaccine calendarcompliance, common signals and symptoms compatible withmercurialism, endemic diseases, suspected or confirmed neuro-pathies and visual or auditory deficits (referred by parents andteachers). Anthropometric data (weight and height) were recordedby two trained nurses, physical examination was conducted by aqualified physician (M.F.F.) and psychometric tests were appliedby two trained psychologists.

All children from both groups were asked and agreed toparticipate; the number of children from Puruzinho (38; 42.1%male) represented all students attending school during the visits.The participating children (32; 43.8% male) from four Iuna ruralschools represent the one which had hair samples collected; therewas no refusal to participate in the study and only analyticalresources (hair-Hg (HHg) analysis) limited the inclusion of morechildren. Local teachers assisted with details in order to promote acloser emotional relationship between children and the researchteam, essential for neurobehavioral evaluation. Scalp hair samples

Fig. 2. Age and school grade distribution of tested children in the Riparian

(Puruzinho) and Agrarian communities (Iuna).

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–1015 1011

were collected from both groups, stored and taken to thelaboratory for total Hg determination.

2.2. Clinical examination

A complete clinical examination was carried out on all testedchildren including an neurological evaluation: postural control andequilibrium (Romberg sign); vibratory sensitivity (utilizing a128 Hz tuning fork); tactile discrimination between two points(normal if below 0.5 mm); walking; deep tendon reflex in threedifferent spinal cord segments (aquiles: sacral S1–S2, patellar:lumbar L2–L4 and finger flexors: cervical/thoracic C6–T1); motorcoordination (rapid alternative and bimanual finger movements);intentional or at rest occurrences of tremor; abnormal eyemovements (asymmetry or nystagmus); muscular strength;stereognosis; joint position sense (hallux); clinical evaluation ofspeech, auditory and visual capacity (without lens).

2.3. Wechsler Intelligence Scale for Children-III (WISC-III)

The WISC-III battery used in the study is the most currentversion validated for the Brazilian population (Figueiredo, 2003;CFP, 2004) and is widely used to evaluate major intellectualabilities of children of 6–17 y of age. Originally, it consisted of 13core subtests (10 elementary and 3 supplementary), whichgenerates three main scales: the Verbal Scale, the PerformanceScale and the Full Scale. The complete Verbal Scale (Verbal-IQ)measured language expression, comprehension, listening, and theability to apply these skills to solving problems; the examiner gavethe questions orally, and the child gave a spoken response. Thecomplete Performance Scale (Performance-IQ) assessed nonverbalproblem solving, perceptual organization, speed, and visual-motorproficiency. Tasks like puzzles, analysis of pictures, imitatingdesigns with blocks, and copying were included in this subtestgroup. Besides, the factors Verbal Comprehension, PerceptualOrganization, Freedom from Distractibility and Processing SpeedIndexes were complementary for estimating specific cognitiveabilities. The Full Scale score (Full Scale-IQ) may be considered anindex of general intellectual functioning.

In this work, seven WISC-III core subtests (whose scores rangefrom 1 to 19) were chosen to assess different aspects of children’sintellectual capacity in both Riparian and Agrarian groups; thechildren were tested and scored by two trained psychologistsaccording to standard procedures. The administered WISC-IIIVerbal subtests and their respective main centers of attention wereSimilarities (SIM; abstract reasoning, categories and relationshipssurvey) and Comprehension (COM; social judgment and commonsense reasoning). Intentionally, three verbal subtests (Information,Arithmetic and Vocabulary) were not used because of their knownsusceptibility to low educational and socioeconomic influences(Figueiredo, 2003). The Performance subtests and their respectivemain centers of attention were: Picture Completion (PC; alertnessto essential detail), Coding (COD; visual-motor coordination, speedand concentration), Picture Arrangement (PA; sequential andlogical thinking), Block Design (BD; spatial and abstract visualproblem solving) and Object Assembly (OA; visual analysis andconstruction of objects).

Given the poor academic environment and low socioeconomiclevels of these populations, we used one of the original mainscales: Performance-IQ (SPER = PC + COD + PA + BD + OA), andone complementary factor: Perceptual Organization Index (PO-i = PC + PA + BD + OA). While SPER score assesses global nonverbalintellectual capacities, PO-i score assesses children’s ability toexamine a problem, draw upon visual-motor and visual-spatialskills, organize their thoughts, create solutions, and then test them.

It can also tap preferences for visual information, comfort withnovel and unexpected situations, or a preference for learning bydoing. Verbal-IQ, Full Scale-IQ and the factors Verbal Comprehen-sion, Freedom from Distractibility and Processing Speed indexeswere not used but both, Verbal subtests sum (SVER = SIM + COM)and Total subtests sum (STOT = SVER + SPER) were applied;STOT was considered as the best estimate of mental abilities.All WISC-III scores were standardized through a percentage scale(0–100) for easier comparisons.

2.4. Human Figure Drawings (HFD)

After finishing WISC-III battery test children were asked todraw the figure of a man with a pencil on a sheet of A-4 paper. HFDis a nonverbal test, mainly based on visual-spatial and construc-tional abilities that may measure children’s intelligence andcognitive maturation. This psychological evaluation and scoredeterminations was done by experienced psychologists with anadapted version for the Brazilian population (CFP, 2004). The HFDpsychometric test is currently used to rank seven ordinarycategories of mental capacity: Deficient (when HFD score below70), Frontier (70–85), Below Medium (85–90), Medium (90–110),Above Medium (110–120), Superior (120–130) and Very Superior(above 130) (Hutz and Bandeira, 2003). Because of its ease ofadministration the HFD Test was used as a complement to theWISC-III.

2.5. MeHg exposure

Hair samples were cut from the occipital area near to the scalp,stored and taken to the laboratory for total Hg determination. Hairchemical analysis was done according to our routine laboratoryprocedures (Bastos et al., 1998) at the Laboratorio de Biogeoquı-mica Ambiental Wolfgang C. Pfeiffer (BIOGEOQ/UNIR) and in theLaboratorio de Radioisotopos Eduardo Penna Franca (LREPF/UFRJ).After mineralization in acid-oxidant media, Hg determination wasperformed with atomic absorption spectometry in a FIMS-4001

PerkinElmer. All analytical runs included certificated material(IAEA-085 and IAEA-086) to assure satisfactory quality control.Recovery rates were above 80% and detection limit below0.05 mg g�1.

2.6. Statistics

Kolmogorov–Smirnov and Shapiro–Wilk tests were used to testthe null hypothesis of normality of distribution. Because normaldistribution was not found for most variables, the Chi-squarestatistic and Mann–Whitney U-test were used to compare the twogroups. Spearman rank correlation (bivariate) was used to testassociation among variables. All significance tests that were two-tailed were considered statistically significant when p < 0.05.Statistical analyses were run with SPSS 15.0 for Windows(www.spss.com, USA).

3. Results

The overall characteristics of children of the two subsistencecommunities are shown in Table 1, while Fig. 3 summarizes theresults of psychometric tests. The fish consumption rates of thecommunities of Riparian and Agrarian children are distinctivelyasymmetric. The contrasting difference between these groups inrespect to MeHg exposure is statistically significant (nonpara-metric Mann–Whitney U-test; p = 0.000); the high median HHg ofRiparians (14.4 mg g�1) contrasts with the low value in Agrarianchildren (0.25 mg g�1). This reflects a subsistence of daily fish

Fig. 3. Psychometric tests scores of children from Puruzinho Lake (Riparians; n = 38) and from Iuna rural families (Agrarians; n = 32); boxplot shows median, percentiles and

outliers; statistical significance in nonparametric Mann–Whitney U-test: n.s.: not significant; WISC-III scores are PC: picture completion; COD: coding; SIM: similarities; PA:

picture arrangement; BD: block design; OA: object assembly; COM: comprehension; SVER: Verbal subtests sum (SIM + COM); SPER: Performance-IQ

(PC + COD + PA + BD + OA); STOT: all subtests sum (SVER + SPER); PO: Perceptual organization (PC + PA + BD + OA); all WISC-III scores were standardized within a

percentage scale; HFD: Human Figure Drawings test.

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–10151012

consumption (80% of Puruzinho villagers eat two fish meals a day)in the Riparian children while Agrarian children of Iuna rarely eatfish or seafood. Despite dissimilar food habits (especially fishconsumption rates) there was no statistically significant differencein anthropometric Z-scores between the two groups. Actually bothpopulations showed signs of mild undernutrition; anthropometricZ-scores were consistently negative (Table 1).

During physical and clinical examination, no salient neurolo-gical alteration was observed in children of the two populations,except one case of nystagmus in Iuna. Nevertheless, there was atendency towards deep tendon hyporeflexia or areflexia (osteo-tendinous reflex) in the Riparian children. In fact, the differencesbetween frequency of occurrence were statistically significant forboth deep tendon reflexes: achilles and patellar reflex arch(nonparametric Pearson Chi-Square test; p = 0.006 and 0.001,respectively). The frequencies of hyporeflexia in finger flexor reflexarch were close to statistical significance (p = 0.052). However, wetested differences in hyporeflexia within fish–MeHg exposedRiparian children: differences in HHg concentrations were notstatistically significant between children with and withoutabnormally hypoactive deep tendon reflex in any sacral (medians10.81 and 14.01 mgHg g�1; p = 0.277), lumbar (medians 11.72 and14.90 mgHg g�1; p = 0.335) and cervical/thoracic (medians 13.44and 14.35 mgHg g�1; p = 0.921) spinal cord segments (nonpara-metric Mann–Whitney U-test).

Clinical neurological examination did not reveal any obviousdeficits compatible with mercurialism in the Riparian group(exposed to fish–MeHg). Both groups performed similarly in theHFD test of cognitive maturation; however, both groups performedpoorly in the psychometric WISC-III test battery (Fig. 3). Medians ofWISC-III subtest scores (STOT) were 17.9 and 28.6 (percentagescale) for Riparian and Agrarian children respectively. Compared tothe Riparian group, the Agrarian children showed a slightly better(but statistically significant) performance in most of the WISC-IIItests; only SIM and OA subsets were not significantly different. The

box-plot in Fig. 3 shows mostly single outliers occurring in bothgroups. On the basis of the nonparametric statistical test used, it issafe to assume that these outliers did not affect the group’sstatistical differences.

Although the Brazilian version of WISC-III test battery has beenstandardized to differences in child age, analyzed together, thesechildren exhibited peculiar responses. Despite low correlationcoefficients, these non-urban children showed significant negativecorrelations between age and WISC-III SIM (r = �0.454; p = 0.000),PA (r = �0.253; p = 0.034), BD (r = �0.342; p = 0.004) and COM(r = �0.292; p = 0.016) subtests, and WISC-III STOT (r = �0.342;p = 0.004), SVER (r = �0.427; p = 0.000), SPER (r = �0.258;p = 0.031) and PO-i (r = �0.278; p = 0.020) factor scores. Despitesignificant negative correlation with age, only the WISC-III subtestSIM-scores was significantly correlated with school level (which isdirectly related to age). However, it is interesting to note thatphysical development was positively correlated with several of theWISC-III scores (which included the one that correlated with age).There were significant correlations between height-for-age Z-scores and following variables of WISC-III: OA subtest (r = 0.269;p = 0.043) and STOT (r = 0.319; p = 0.016), SPER (r = 0.311;p = 0.019) and PO-i (r = 0.302; p = 0.023) (Table 2).

4. Discussion

Although the mean HHg concentration of the Riparian groupwas 66 times higher than in the Agrarian group, the median STOTscores (WISC-III general intellectual functioning) were clearlybelow 50 (percentage scale) in both populations; the Agrarianchildren performed slightly (but significantly) better than theRiparians. Because of the very low scores in both groups, thesefindings are difficult to interpret. Difference in fish consumptionalone (and attendant MeHg exposure) is not sufficient to explainsuch poor neurocognitive performance in isolated communitieswith a low academic environment. We recognize two limitations of

Table 2Summary of nonparametric correlation (Spearman’s) analysis between developmental antropometric and psychometric variables of 70 children from both groups (Riparians,

38; and Agrarians, 32)

Psychometric variables Age

(years)

School level

(years)

Height-for-age and sex

(Z-score)

Weight-for-age and sex

(Z-score)

BMI-for-age and sex

(Z-score)

PC

r �0.169 0.197 0.259 0.163 0.077

p 0.162 0.103 0.052 0.217 0.569

COD

r �0.097 0.140 0.205 0.119 0.017

p 0.422 0.247 0.126 0.371 0.900

SIM

r �0.454** �0.272* �0.046 �0.235 �0.216

p 0.000 0.024 0.734 0.075 0.109

PA

r �0.253* 0.054 0.157 0.049 0.036

p 0.034 0.657 0.245 0.711 0.790

BD

r �0.342** 0.067 0.256 0.030 �0.151

p 0.004 0.579 0.055 0.821 0.261

OA

r �0.218 0.126 0.269* 0.136 0.021

p 0.070 0.297 0.043 0.305 0.877

COM

r �0.292* �0.082 0.226 0.032 �0.124

p 0.016 0.508 0.090 0.813 0.357

STOT

r �0.342** 0.042 0.319* 0.092 �0.096

p 0.004 0.734 0.016 0.490 0.478

SVER

r �0.427** �0.179 0.164 �0.090 �0.244

p 0.000 0.145 0.222 0.498 0.067

SPER

r �0.258* 0.141 0.311* 0.145 �0.006

p 0.031 0.244 0.019 0.275 0.964

PO-i

r �0.278* 0.146 0.302* 0.126 �0.003

p 0.020 0.228 0.023 0.343 0.985

HFD

r �0.158 0.178 0.243 0.057 �0.099

p 0.198 0.147 0.071 0.670 0.468

**Statistically significant at the 0.01 level (2-tailed); *Statistically significant at the 0.05 level (2-tailed); WISC-III scores are PC: picture completion; COD: coding; SIM:

similarities; PA: picture arrangement; BD: block design; OA: object assembly; COM: comprehension; SVER: Verbal subtests sum (SIM + COM); SPER: Performance-IQ

(PC + COD + PA + BD + OA); STOT: all subtests sum (SVER + SPER); PO-i: Perceptual organization index (PC + PA + BD + OA); all WISC-III scores were standardized within a

percentage scale; HFD: Human Figure Drawings test.

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–1015 1013

the study: (a) the small group size and (b) other alternative factorsthat might affect development in these different populations.

Neurodevelopmental tests among children growing up in poorenvironments are known to be modulated by related stressors:poverty, malnutrition and parasitemia leading to iron deficiencyanemia (Gordon, 2003). Inasmuch as children’s nutritional status isreflected by growth, others have found low performance inneurocognitive tests among smaller children of underdevelopedcountries (Klein et al., 1972; Freeman et al., 1977; Lasky et al.,1981; Bogin and MacVean, 1983); in these children, inadequatenutrition seems to result not only from limited food availability butalso unhealthy sanitary conditions which increase morbidity.Current studies have estimated that more than one-third ofAmazonian riparian children are iron deficient; malnutrition andanemia are significantly more prevalent among children from alower socioeconomic stratum (Muniz et al., 2007). Therefore, itwas surprising to find out that cognitive scores could also benegatively correlated with age (all children; Table 2). The mostrecent review on human developmental neurotoxicity of MeHg

covering all pertinent studies does not mention age relatedconfounders (Castoldi et al., 2008); in the studies reviewed the agespan was much narrower than in our study.

Overall, our results concur with a recent study in youngerriverine children of the Alto Pantanal (Rio Cuiaba, BrazilianAmazon). Tavares et al. (2005) reported that in both control groupand riverine children, there was a high proportion of them withwhat was considered ‘‘non-normal’’ performance; Tavares et al.(2005) suggested that the low performance (in the LefreveEvolutional Neurological Test) could not be related to mercuryexposure and warned that ‘‘this type of test presented limitations for

river-dwelling Amazon children.’’ It is important to point out thecultural similarities between the groups studied by Tavares et al.(2005): they were both from the same region but the control grouphad a more varied diet that ended up with a mean HHg (2.08 ppm)half that of the more exposed group (5.37 ppm). Compared to ourstudy, the group difference in mean HHg is much higher (66 times)than the small difference (two times) reported between groups inthe Tavares et al. (2005) study. The levels of fish consumption of

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–10151014

the Riparian children are high but close to children of both riverineand Amer-Indian children of comparable age. We have reported amean of 18.5 mg g�1 of Hg for children of the Rio Negro (Barbosaet al., 2001) and 16.5 mg g�1 of Hg for Amer-Indian children ofcommunities of the Rio Tapajos (Dorea et al., 2005).

Grandjean et al. (1999) studied 7 and 12 y old children in threeTapajos villages with fish–MeMg exposures lower than ourRiparian group (mean HHg, 11 mg g�1); they claimed thatdecrements in neuropsychological tests of motor function,attention, and visual-spatial performance were associated withHHg concentrations. In that study, a medical student administeredthe motor function tests, while a nurse administered neuropsy-chological tests. The version used of these tests had not beenvalidated for the Brazilian populations (tests were given inPortuguese but included an interpreter only in one of the moreexposed populations). Although socioeconomic conditions werereferred to as similar among studied populations, the moreexposed villages were also the least impacted by deforestation,automobile traffic and industrialization (parameters of geographicisolation). Indeed, a spatial trend of increasing HHg concentrationas a function of village distance from an urban area has been shownfor Amazonian riverines (Alves et al., 2006).

Cordier et al. (2002) studied Ameri-Indian children of FrenchGuyana, comparing low with high-fish–MeHg exposure and foundno significant association between children’s HHg and neurode-velopmental parameters. However, they reported an inverse dose-dependent association between maternal HHg and decreasedperformance in visual-spatial organization (Stanford-Binet Copy-ing score). Other neuromotor alterations reported by Cordier et al.(2002) included hyperactive deep tendon reflexes and poorercoordination of the legs. However, McKeown-Eyssen et al. (1983)reported that abnormal muscle tone and tendon reflexes werepresent only in Cree Indian (Canada) young boys and showed nodose–response relationship with maternal HHg (mean concentra-tion, 6 mg g�1); they considered these findings of questionableclinical significance and concluded that the data may represent achance occurrence. Indeed the hyporeflexia that we observed inthe children is not associated with any kind of motor dysfunctionnot observed in the adult population. Others have testedneurobehavioral batteries in Brazilian Amazonian communitieswith different results.

According to UNICEF (1990), 35% of the Brazilian populationis comprised of school-age children with nearly 50% still livingbelow the poverty line. These adverse conditions reflect directlyand indirectly in physical and cognitive development ofyoungsters: reduced stature and first grade failure rates reach70% in urban-slum communities (Paine et al., 1999). Thechildren in our study groups are exposed to intestinal parasites,but the Puruzinho sample were additionally exposed to endemicmalaria which aggravates anemia because splenic hemolysis issecondary to protozoa-infected red cells. Another feature thatmight contribute to poor performance is the deficit in stimulidue to isolation, which is common to both groups but muchmore accentuated in the Riparians. Although the group medianswere not statistically significant for growth differences (Z-score), the significant correlation between height-for-age Z-score and intellectual performance is in agreement with otherstudies (Paine et al., 1999). Indeed physical development waspositively correlated with several of the WISC-III scores (whichincluded the ones that correlated with age). This, partly, mayaccount for the significant correlation between age andcognitive performance. The recent review of Castoldi et al.(2008) summarized the most important studies of neurocogni-tive tests related to fish–MeHg exposure showing very fewstudies that could match the age range of our children and none

had studied growth and/or attendant age differences in relationto cognitive performance.

Fish–MeHg exposure during intrauterine growth is a debatableissue with divergent neurocognitive outcomes: while no signifi-cant effect has been reported for the Seychelles cohort, significantdifferences have been reported for both Faroe Islands and the NewZealand studies; it is worth mentioning that Hg concentration inseafood varies from an average of 0.3 ppm in ocean fish consumedin the Seychelles Republic, up to 3 ppm in whale meat on the FaroeIslands, and to as much as 4 ppm in shark consumed in NewZealand (van Wijngaarden et al., 2006). Besides differences inintrauterine MeHg exposure other environmental differences areemerging. Huang et al. (2007), reanalyzing the data from theSeychelles cohort, identified subgroups of children that respondednon-homogenously to fish consumption during pregnancy; theyreported that ‘‘motor proficiency significantly decreased with

increasing prenatal MeHg exposure in 7% of the children whose home

environment was below average.’’ These findings, along with ourobservation of a negatively correlation between psycometricscores and age point to confounders, that can shift the debate ofneurocognitive adverse effects from exclusive fish–MeHg exposureto other environmental variables.

5. Conclusion

The high dissimilarity of fish–MeHg exposure between Riparianand Agrarian children can not explain comparable poor neurode-velopment performance, possibly a slightly better academicenvironment in Iuna could account for the better performanceof Agrarian children. Global strategies for reducing humanexposure to all forms of Hg are desirable but recommendationsto curtail fish consumption are unreal decisions for riverinesubsistence populations and are not justifiable to preventdiminished neurocognitive scores.

Acknowledgments

We thank the staff and children of the Puruzinho and Iunaschools for their cooperation and the Psychologists RaysonMarques, Verusca Santos, Prof. Valeria Oliveira and MarciaRodrigues for the psychometric tests. We are also in debt to thestaffs of Iuna Municipality and the Faculty Porto Seguro. Dr. J.P.M.Torres is Advanced Selikoff Fellows at the Mount Sinai School ofMedicine and Queens College in New York and is supported in partby grant 1 d43 tw00640 from the Fogarty International Center ofthe National Institute of Health (NIH).

Conflict of interest

We declare no conflict of interest.

References

Alves MFA, Fraiji NA, Barbosa AC, Lima DSS, Souza JR, Dorea JG, Cordeiro GWO. Fishconsumption, mercury exposure and serum antinuclear antibody in Amazonians.Int J Environ Health Res 2006;16:255–62.

Axelrad DA, Bellinger DC, Ryan LM, Woodruff TJ. Dose–response relationship ofprenatal mercury exposure and IQ: an integrative analysis of epidemiologic data.Environ Health Perspect 2007;115:609–15.

Barbosa AC, Jardim W, Dorea JG, Fosberg B, Souza J. Hair mercury speciation as afunction of gender, age, and body mass index in inhabitants of the Negro Riverbasin, Amazon, Brazil. Arch Environ Contam Toxicol 2001;40:439–44.

Barbosa AC, Silva SR, Dorea JG. Concentration of mercury in hair of indigenous mothersand infants from the Amazon basin. Arch Environ Contam Toxicol 1998;34:100–5.

Bastos WR, Gomes JP, Oliveira RC, Almeida R, Nascimento EL, Bernardi JV, de LacerdaLD, da Silveira EG, Pfeiffer WC. Mercury in the environment and riversidepopulation in the Madeira River Basin, Amazon, Brazil. Sci Total Environ2006;368:344–51.

M.F. Fonseca et al. / NeuroToxicology 29 (2008) 1008–1015 1015

Bastos WR, Malm O, Pfeiffer WC, Cleary D. Establishment and analytical quality controlof laboratories for Hg determination in biological and geological samples in theAmazon—Brazil. Ciencia e Cultura. J Braz Assoc Adv Sci 1998;50:255–60.

Bogin B, MacVean RB. The relationship of socioeconomic status and sex to body size,skeletal maturation, and cognitive status of Guatemala City schoolchildren. ChildDev 1983;54:115–28.

Castoldi AF, Johansson C, Onishchenko N, Coccini T, Roda E, Vahter M, et al. Humandevelopmental neurotoxicity of methylmercury: impact of variables and riskmodifiers. Regul Toxicol Pharmacol 2008;51:201–14.

CFP (Conselho Federal de Psicologia). Sistema de Avaliacao de Testes Psicologicos(www.pol.org.br). Suplemento Especial de Fevereiro 2004.

Cohen JT, Bellinger DC, Shaywitz BA. A quantitative analysis of prenatal methylmercury exposure and cognitive development. Am J Prev Med 2005;29:353–65.

Cordier S, Garel M, Mandereau L, Morcel H, Doineau P, Gosme-Seguret S, et al.Neurodevelopmental investigations among MeHg-exposed children in FrenchGuiana. Environ Res 2002;89:1–11.

Dorea J, Barbosa AC, Ferrari I, de Souza JR. Mercury in hair and in fish consumed byRiparian women of the Rio Negro, Amazon, Brazil. Int J Environ Health Res2003;13:239–48.

Dorea JG. Fish are central in the diet of Amazonian riparians: should we worry abouttheir mercury concentrations? Environ Res 2003;92:232–44.

Dorea JG. Cassava cyanogens and fish mercury are high but safely consumed in the dietof native Amazonians. Ecotoxicol Environ Saf 2004;57:248–56.

Dorea JG, Barbosa AC, Ferrari I, Souza JR. Fish consumption (hair mercury) andnutritional status of Amazonian Amer-Indian children. Am J Hum Biol2005;17:507–14.

Figueiredo VLM. WISC-III. In: Cunha JA, editor. Psicodiagnostico-V. 5 Edicao Revisada eAmpliada. 3 Reimpressao. Artmed; 2003, p. 603.

Freeman HE, Klein RE, Kagan J, Yarbrough C. Relations between nutrition and cognitionin rural Guatemala. Am J Public Health 1977;67:233–9.

Gordon N. Iron deficiency and the intellect. Brain Develop 2003;25:3–8.Grandjean P, Weihe P, White RF, Debes F, Araki S, Yokoyama R, et al. Cognitive deficits

in 7-year-old children with prenatal exposure to methylmercury. NeurotoxicolTeratol 1997;19:417–28.

Grandjean P, White RF, Nielsen A, Cleary D, de Oliveira Santos EC. Methylmercuryneurotoxicity in Amazonian children downstream from gold mining. EnvironHealth Perspect 1999;107:587–91.

Huang LS, Myers GJ, Davidson PW, Cox C, Xiao F, Thurston SW, et al. Is susceptibility toprenatal methylmercury exposure from fish consumption non-homogeneous?Tree-structured analysis for the Seychelles child development study Neurotox-icology 2007;28:1237–44.

Hutz CS, Bandeira DR. Desenho da Figura Humana. Cap 33. In: Cunha JA, editor.Psicodiagnostico-V. 5 Edicao Revisada e Ampliada. 3 Reimpressao. Artmed;2003, p. 507.

Klein RE, Kagan J, Freeman HE, Yarbrough C, Habicht JP. Is big smart? The relation ofgrowth to cognition J Health Soc Behav 1972;13:219–25.

Lasky RE, Klein RE, Yarbrough C, Engle PL, Lechtig A, Martorell R. The relationshipbetween physical growth and infant behavioral development in rural Guatemala.Child Dev 1981;52:219–26.

McKeown-Eyssen GE, Ruedy J, Neims A. Methyl mercury exposure in northern Quebec.II. Neurologic findings in children. Am J Epidemiol 1983;118:470–9.

Muniz PT, Castro TG, Araujo TS, Nunes NB, Nunes MS, Hoffmann EHE, et al. Child healthand nutrition in the western Brazilian Amazon: population-based surveys in twocounties in Acre State. Cad Saude Pub 2007;23:1283–93.

Paine P, Dorea JG, Pasquali L, Monteiro AL. Growth and cognition in Brazilian school-children: a spontaneous occurring intervention study. Int J Behav Dev 1999;15:169–83.

Palmer RF, Blanchard S, Stein Z, Mandell D, Miller C. Environmental mercury release,special education rates, and autism disorder: an ecological study of Texas. HealthPlace 2006;12:203–9.

Passos CJ, Mergler D, Fillion M, Lemire M, Mertens F, Guimaraes JR, et al. Epidemiologicconfirmation that fruit consumption influences mercury exposure in ripariancommunities in the Brazilian Amazon. Environ Res 2007;105:183–93.

Tavares LM, Camara VM, Malm O, Santos EC. Performance on neurological develop-ment tests by riverine children with moderate mercury exposure in Amazonia,Brazil. Cad Saude Pub 2005;21:1160–7.

Trasande L, Landrigan PJ, Schechter C. Public health and economic consequences ofmethyl mercury toxicity to the developing brain. Environ Health Perspect2005;113:590–6.

UNICEF. Situacao Mundial da Infancia. Brasılia: UNICEF; 1990.van Wijngaarden E, Beck C, Shamlaye CF, Cernichiari E, Davidson PW, Myers GJ,

Clarkson TW. Benchmark concentrations for methyl mercury obtained from the9-year follow-up of the Seychelles child development study. Neurotoxicology2006;27:702–9.