- Email: [email protected]
Environmental exposure to polychlorinated biphenyls and quality of the home environment: effects on psychodevelopment in early childhood
ARTICLES
Environmental exposure to polychlorinated biphenyls and quality of the home environment: effects on psychodevelopment in early childhood Jens Walkowiak, Jörg-A Wiener, Annemarie Fastabend, Birger Heinzow, Ursula Krämer, Eberhard Schmidt, Hans-J Steingrüber, Sabine Wundram, Gerhard Winneke
Summary Background There is uncertainty whether environmental levels of exposure to polychlorinated biphenyls (PCBs) adversely affect mental and motor development in early childhood. We aimed to establish whether such an effect is of only prenatal or additional postnatal origin, and if a favourable home environment can counteract this effect. Methods Between 1993 and 1995 we recruited 171 healthy mother-infant pairs and prospectively measured psychodevelopment in newborn infants aged 7, 18, 30, and 42 months. We estimated prenatal and perinatal PCB exposure of newborn babies in cord blood and maternal milk. At 42 months we measured postnatal PCB concentrations in serum. At 18 months the quality of the home environment was assessed using the Home Observation for Measurement of the Environment scale. Mental and psychomotor development of the children were assessed using the Bayley Scales of Infant Development until 30 months and the Kaufman Assessment Battery for Children at 42 months. Findings Negative associations between milk PCB and mental/motor development were reported at all ages, becoming significant from 30 months onwards. Over 30 months, for a PCB increase from 173 (5th percentile) to 679 ng/g lipids in milk (95th percentile) there was a decrease of 8·3 points (95% CI –16·5 to 0·0) in the Bayley Scales of Infant Development mental scores, and a 9·1 point decrease (95% CI –17·2 to –1·02) in the Bayley Scales of Infant Development motor scores. There was also a negative effect of postnatal PCB exposure via breastfeeding at 42 months. Home environment had a positive effect from 30 months onwards (Bayley Scales of Infant Development mental score increase of 9·4 points [95% CI 2·2–16·7]). Interpretation Prenatal PCB exposure at current European background levels inhibits, and a favourable home environment supports, mental and motor development until 42 months of age. PCB exposure also has an effect postnatally. Lancet 2001; 358: 1602–07 See Commentary page 1568 Divisions of Analytical Chemistry, Epidemiology, and Neurobehavioral Toxicology at Medical Institute of Environmental Hygiene (A Fastabend PhD, U Krämer PhD, J Walkowiak PhD, Prof G Winneke PhD, S Wundram), and Institute of Medical Psychology (J A Wiener, Prof H J Steingrüber PhD), and University Child Clinic, Medical Faculty (Prof E Schmidt MD), Heinrich-HeineUniversity Düsseldorf; and Laboratory of Environmental Toxicology at the State Agency of Nature and Environment of SchleswigHolstein, Flintbeck, Germany (B Heinzow MD) Correspondence to: Dr Gerhard Winneke, Medical Institute of Environmental Hygiene at Heinrich-Heine-Universität Düsseldorf, Auf’m Hennekamp 50, D-40225 Düsseldorf, Germany (e-mail: [email protected])
1602
Introduction Both endogenous and exogenous factors and their interactions regulate human child development. Among the endogenous factors, genetic and biochemical induction and modulation are predominant, whereas psychosocial and physicochemical factors and their interactions are important exogenous determinants of general development or psychodevelopment of the human infant. As for the psychosocial dimension many different individual aspects of the child environment have received attention in developmental studies.1 However, the isolated study of such variables underestimates the contribution of the many interacting environmental components for development. Therefore, more comprehensive constructs for a quantitative description of relevant features of home environments have been developed. A prominent instrument in this respect is the Home Observation for Measurement of the Environment (HOME).2 We have used the score to compare the developmental effects of the quality of the home environment with those of early chemical exposure. Among the exogenous factors affecting early neurobehavioural child development is the exposure to chemicals in utero or during the early stages of brain development.3 Some environmental chemicals, such as inorganic lead or organic mercury, have received attention in this respect, including the polychlorinated biphenyls (PCBs). PCBs are persistent environmental contaminants consisting of up to 209 individual congeners. PCBs and their metabolites cross the placenta, exposing the vulnerable fetus to PCBs circulating in maternal blood. After birth the infant is additionally exposed to relatively high PCB concentrations in human milk. Among a broad spectrum of biological effects, developmental neurotoxicity seems to be a prominent feature of these chemical mixtures, as can be judged from findings in animals and human beings.4 Evidence of neurodevelopmental effects of prenatal or early postnatal PCB exposure in human beings at environmental exposure levels is largely based on findings from four sets of cohort studies.5–10 Although all of these studies did report some adverse effects of early PCB exposure on neurological or cognitive development, they are not fully consistent with regard to confounding, spectrum, persistence of effects, or the biological matrix in which PCBs were reported most predictive for later developmental effects.11 Previous reports describing neurodevelopmental effects of environmental chemicals have typically concentrated on exposure-response effects in isolation, and have treated other factors of developmental impact merely as confounders. We have compared the independent effects of early PCB-exposure and the home environment from the age of 7–42 months. Part of the findings at 7 months has already been reported;12 they are covered again in a modified manner here, to allow for direct comparisons with the observations made at the later ages.
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Methods Study population We recruited 171 healthy mother-infant pairs between October, 1993, and May, 1995, from the obstetrical wards of three Düsseldorf hospitals. Three advanced medical students were involved in the recruitment process. Criteria for inclusion were: principal agreement of the mother, first and second born babies delivered at term (weeks 37–42 of pregnancy) from native German families; an Apgar score at 5 minutes of at least 78; and no serious illnesses or complications during pregnancy and delivery. The restriction to parities one and two served to keep the sample more homogeneous with respect to PCB exposure. Those fulfilling these criteria and having an appropriate medical condition were approached for participation, and about 70% of these, with written consent, agreed to participate until their babies were at least 7 months of age. The study protocol was approved by the medical ethics committee of the Heinrich-HeineUniversity, Düsseldorf, Medical Faculty. Quality of the home environment The HOME score by Caldwell and Bradley2 is a powerful and validated instrument for the assessment of the home environment both as a developmental determinant and an influential confounder in neurotoxicity studies.13–15 The infant version (0–3 years) used here consists of 45 items covering emotional and verbal responsivity of the mother, acceptance of the child’s behaviour, organisation of the home environment, availability of toys, parental involvement with the child, and variability of daily experience. Assessment takes place during a semistructured interview with the mother at home in the presence of the child. We used a version adapted to fit the German cultural context16 at 18 months of age. Retest reliabilities, checked by two observers in an independent sample of 26 children at ages 5 and 16 months (observer A) as well as ages 7 and 18 months (observer B), were rtt=0·71–0·72 for these two 11 month intervals.16 PCB exposure We used the sum of the three PCB congeners 138, 153, and 180 as the marker of PCB exposure. We took cord blood samples in polyethylene monovettes and stored them at –20˚C until analysis according to Fastabend.17 At 42 months we drew venous blood (5 mL) using vacutainer glass tubes for the separation of serum. Blood samples were allowed to clot and then centrifuged at 3000 rpm. We transferred the serum into glass vials and stored them at –20˚C until sample preparation and analysis. For PCB analyses in human milk spot samples (5–10 mL) were collected around 2 weeks postpartum in glass vials, and stored at –20˚C until analysis according to a standard method.18 Briefly, after thawing and homogenisation, we extracted the PCBs by n-heptane and purified the extracts by a silica gel column. We did PCB analyses on a highresolution gas chromatograph with electron capture detection. Three marker congeners 138, 153, and 180 were determined with a detection limit of 0·01 µg/L (serum) and 0·005 mg/kg fat (milk). We determined the lipid content with the photometric Mercko-Test 3221 (Merck, Darmstadt). Psychodevelopmental tests Psychomotor and cognitive development were assessed at 7, 18, 30, and 42 months. Testing was done at the child’s home, usually in the presence of the mother, by the same examiner (JAW), who was unaware of the child’s PCB
THE LANCET • Vol 358 • November 10, 2001
concentration, for the whole study period within 2 weeks of the target age. Of the Bayley scales of Infant Development, an established psychodevelopmental tool, the mental and the motor scales were used. The mental scale assesses the child’s level of cognitive functioning, language development, and personal/social development. The motor scale assesses fine, and gross motor functioning. For reasons of comparability raw scores were transformed to yield a mean of 100 and an SD of 15. At 42 months of age, we measured intelligence with the German version of the Kaufman Assessment Battery for Children.19 Scores from the Sequential Processing and Simultaneous Processing subscales are combined to yield the Mental Processing Composite-Index which is standardised to a mean of 100 and an SD of 15. Potential confounders/covariates Potential confounders included age at examination (days), gestational age (weeks), alcohol/smoking during pregnancy (separately scored as 0/1), Apgar score (8, 9, 10), neonatal illness/jaundice (0/1), spontaneous delivery (0/1), parity (1/2), lead in cord blood (µg/L), chronic diseases during past year (0/1), duration of breastfeeding (months), parental education (years of schooling; maximum of either parent), parental occupation (10 categories;20 highest of either parent), body-mass index of mother (kg/m2), and maternal verbal intelligence quotient (subtest vocabulary from the German form of the Wechsler Adult Intelligence Scale).21 Statistical analysis We treated the data both descriptively as well as inferentially with SAS (version 8.01). To analyse the association of HOME and PCBs with the target variables we used multiple linear regression. Model building involved two steps: in the first step, variables were selected based on theoretical knowledge of their effect on mental or motor development, or on empirical evidence from previous studies on the relation of either PCBs or the HOME to the respective outcome variables. In the second step, selection of variables was data-driven; only those variables were selected which had correlations at p<0·20 with both the exposure index and at least two of the outcome variables after inclusion of the first step variables. The variables in the first step were parental education, sex, maternal intelligence quotient, as well as HOME and PCB simultaneously. Those in the second step were parity, smoking in pregnancy, and body-mass index. All outcome variables were uniformly adjusted for these variables. Interactions were considered but discarded if not significant. Additionally, to take advantage of the time-series aspect of the Bayley Scales of Infant Development administered at three points in time, a repeated measurements analysis was also done. We used one-tailed probabilities throughout, because the hypotheses were directional, and effects at p<0·05 were judged significant.
Results From recruitment until 42 months of age the sample size dropped from an initial 171 to 116. 126 mothers from the initial 171 provided milk samples at 2 weeks and 91 of these remained in the study until final testing at 42 months. Despite this attrition the sample structure changed only marginally. No significant difference from recruitment to testing at 42 months was detected for PCB concentrations, sex, birthweight, gestational/maternal age, neonatal condition, parental education, maternal
1603
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
PCB concentrations
Percentiles n*
Cord PCB (ng/mL) 141 Milk PCB (ng/g lipids) 126 Serum PCB at 42 92 months (ng/mL)
5%
median 75%
95%
0·11 0·28 0·39 0·50 0·83 173 294 404 535 679 0·23 0·49 1·22 1·87 3·35
Duration of breastfeeding (PCB ⭐2 weeks (ng/mL) 15 >2 weeks to 4 months 32 (ng/mL) >4 months (ng/mL) 44 HOME (total score)† Sum of six subscales 156
25%
42 months)* 0·14 0·24 0·31 0·49
0·36 0·68
0·52 1·20
4·39 1·79
0·29
1·77
2·50
3·52
34
1·37 38
41
42
44
Number of breastfed children is 91, one child with missing data on duration of nursing. *Different sample sizes due to incomplete information. †Raw-scores of the HOME scale administered at 18 months. The theoretical maximum is 45.
Table 1: Cord, milk and serum PCB concentrations at different ages and relative to breastfeeding, and HOME scores
intelligence quotient, neonatal illness, alcohol/smoking in pregnancy, and lead in cord blood. However, there was a tendency for families staying within the study to be better educated. Table 1 shows that PCB concentrations in serum samples at 42 months increased markedly with duration of breastfeeding and were about five times higher than those in the group of non-breastfed children. Children with a breastfeeding period of 2 weeks or less were included in the group of non-breastfed children. Because the PCB distributions were skewed all subsequent analyses are based on log2-transformed values. The correlations between PCB in cord blood and milk were positive and highly significant (r=0·57; p<0·0001). The range of HOME scores was quite narrow and were shifted to the upper end of the environmental quality scale (table 1). Mean maternal age (SD) was 29·5 years (4·4), mean gestational age was 39·9 weeks (1·2), and the average birthweight was 3493 g (463). Lead concentrations in cord blood were very low (20·3 [7·1] µg/L). Mean raw scores of the Bayley Scales of Infant Development mental scale increased from 64·5 (SD 2·2) at 7 months to 149·3 (5·1) at age 30 months. For the Bayley Scales of Infant Development motor scale this increase was from 40·7 (3·5) at 7 months to 94·8 (2·9) at 30 months. For the Kaufman Assessment Battery for Children at 42 months the mean score was 103·5 (10·9). Table 2 shows results from the regression analyses relating PCB concentrations in maternal milk and the HOME scale to mental and motor development. After adjustment, exposure-response associations with PCB were negative and of only borderline significance at ages 7 and 18 months, but became significant at 30 (Bayley Scales of Infant Development) and at 42 months of age (Kaufman Assessment Battery for Children). Treating the age-range of 7–30 months as a time-series, early PCB exposure shows significant negative overall associations
with mental/motor development. Only children with complete data were included. However, the strength of associations did not change when using simple imputation techniques to account for missing values. Negative associations with PCB increased with age, and positive associations with the quality of the home environment became more pronounced at 30 or 42 months of age. Dose-response associations at 30 and 42 months of age, based on adjusted values for the Bayley Scales of Infant Development mental/motor development and the Mental Processing Composite-Index of the Kaufman Assessment Battery for Children, reveal a pattern of facilitation for the home environment and one of inhibition for milk PCB with rather similar slopes for both variables over the range of observed HOME scores or milk PCB concentrations (figure 1). Yet, the overall positive impact of the home environment on development is stronger than the negative effect of neonatal PCB exposure. This is evident from figure 2 which illustrates effectsizes derived from the respective regression coefficients from table 2. Here, the changes of the Bayley Scales of Infant Development scores or Kaufman Assessment Battery for Children scores are given for an increase of HOME scores or PCB concentrations from the lower 5th to the upper 95th of their distributions from table 1. To exemplify for the 30-months assessment of Bayley Scales of Infant Development mental score, the increase with increasing HOME score was 17·7, whereas the decrease with increasing milk PCB concentrations was just 9·9 points. However, for the time-series approach covering the full 7–30 months period (not given in figure 2) effect-size differences between HOME and milk PCB concentrations were less pronounced: the overall increase for Bayley Scales of Infant Development (mental) with increasing HOME score was 9·4 points (95% CI 2·2–16·4), and the overall decrease with increasing PCB concentraion was –8·3 points (–16·5 to 0·0); a very similar pattern emerged for Bayley Scales of Infant Development motor development: an increase with increasing HOME of 9·8 points (2·7–16·9) and a decrease with increasing PCB concentrations of –9·1 (–17·2 to –1·02). All associations with cord blood PCB and the Bayley Scales of Infant Development mental score were small and even slightly positive. Association between the Bayley Scales of Infant Development motor score and PCB was negative and very small. For example, for the time-series model and using the uniform confounder model the final outcome was as follows: t1,98=–0·93, p=0·36. The associations between cord blood PCB and the Mental Processing Composite-Index of the Kaufman Assessment Battery for Children at 42 months were small and even slightly positive. The possibility of an additional developmental impact of postnatal PCB intake which, at this age, occurs mainly through breastfeeding (table 1), was tested by means of two models. The first model used a rough estimate of
Mental development PCBmilk log2 [ng/g]
7 months (BSID; n=110) 18 months (BSID; n=112) 30 months (BSID; n=104) 7–30 months (cohort analysis; n=104) 42 months (K-ABC; n=87)
Motor development HOME score
PCBmilk log2 [ng/g]
HOME score
*
t
p†
*
t
p†
*
t
p†
*
t
p†
–3·61 –4·11 –4·98 –4·19
–1·26 –1·56 –1·80 –1·99
0·10 0·06 0·035 0·025
0·56 0·65 1·77 0·94
1·14 1·47 3·74 2·58
0·13 0·07 0·0002 0·005
–3·13 –4·78 –4·73 –4·61
–1·19 –1·71 –1·68 –2·22
0·12 0·045 0·05 0·015
0·47 0·88 1·82 0·98
1·04 1·88 3·77 2·70
0·15 0·03 0·0002 0·004
–4·30
–1·93
0·028
1·29
3·35
0·0006
··
··
··
··
··
··
K-ABC=Kaufman Achievement Battery for Children. BSID=Bayley Scales of Infant Development. *Regression coefficients. † One-tailed probabilities.
Table 2: Association of PCR and HOME with mental and motor development
1604
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Bayley (standardised values)
Mental development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <38
38–39
40–41
42
>42
Bayley (standardised values)
ARTICLES
Mental development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <276
Bayley (standardised values)
Motor development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <38
38–39
40–41
42
>42
Motor development (Bayley) measured at 30 months
<276
42
>42
Bayley (standardised values)
Kaufman (standardised values)
40–41
>276–352 >352–448 >448–558
>558
Milk PCBs (ng/g lipids)
Mental development (Kaufman) measured at 42 months 114 108 102 96 90 84 78 38–39
>558
114 108 102 96 90 84 78
HOME scale (raw scores)
<38
>276–352 >352–448 >448–558
Milk PCBs (ng/g lipids) Bayley (standardised values)
HOME scale (raw scores)
Mental development (Kaufman) measured at 42 months 114 108 102 96 90 84 78 <276
>276–352 >352–448 >448–558
>558
Milk PCBs (ng/g lipids)
HOME scale (raw scores)
Figure 1: Exposure-response associations for HOME scores or milk PCB and mental or psychomotor development Dose-response associations for Bayley Scales of Infant Development mental and motor scores and Kaufman Assessment Battery for Children scores in relation to quality of the home environment and PCB concentrations (quintiles). Values are means (95% CI) and regression lines are superimposed.
Mental HOME PCBs (milk)
Psychomotor HOME PCBs (milk) K–ABC
BSID
20
‡
‡
Change in score
15
‡
10
† *
5
dose, namely (milk PCB⫻months of breastfeeding) adjusted for prenatal or perinatal exposure. This model was applied to the repeated measurement approach of Bayley Scales of Infant Development (mental and motor) and to the Kaufman Assessment Battery for Children (Mental Processing Composite-Index). The second model was applicable to the Kaufman Assessment Battery for Children as the outcome variable only, and used PCB in 42-months serum, again adjusted for prenatal or perinatal exposure. No significant or borderline association with calculated postnatal PCB exposure was reported for both Bayley Scales of Infant Development measures. However, a significant negative effect of postnatal exposure was noted for both models as applied to the Kaufman Assessment Battery for Children at 42 months of age (t=–1·89, p=0·031 for dose, and t=–2·01, p=0·025 for PCB at 42 months).
0
Discussion –5 *
–10 7 months
†
† †
18 months
30 months
†
42 months
Figure 2: Adjusted effect sizes for mental and psychomotor development in relation to the quality of the home environment or milk PCB concentrations Calculated from the regression coefficients given in table 2 for a change in the independent variables from the 5th–95th percentiles (table 1).*p<0·10; †p<0·05; ‡p<0·001 (one-tailed). Upward columns represent HOME-related developmental facilitation, downward columns represent PCB-related developmental delay or deficit.
THE LANCET • Vol 358 • November 10, 2001
Despite considerable attrition of the cohort during the study period, mainly due to loss of interest, moving away from Düsseldorf, or some cases of initially unidentified congenital disorders, we observed a strong positive effect of the home environment as assessed by the HOME scale for psychodevelopment. With increasing age the degree of association between HOME and mental or motor development increased, an observation which is consistent with published evidence.22 This effect might reflect the extended effect of the quality of the child’s environment, but could also be due to the increase with age of the psychometric quality of developmental testing. Although
1605
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
the range of the HOME scores was small and restricted to the upper end of the continuum, the size of the facilitating effect of the home environment on psychodevelopment was remarkable. Despite their ban in terms of production and use between 1977 and 1983, PCBs are still of environmental concern. The 209 possible PCB congeners differ by the number and position of chlorine atoms on the two benzene rings making up the basic structure of the molecule. From the six marker congeners (28, 52, 101, 138, 153, 180), typically measured to characterise PCB concentrations in environmental media, we only used the three higher chlorinated compounds as indicators of the internal PCB exposure. This is in accordance with recommendations23 and also considers the fact that these three account for about 60% of the total PCB content in human tissue samples.24 Having defined prenatal and neonatal PCB exposure in this manner we have shown that mental and motor development between 7 and 42 months of age has a significant negative association with PCB concentrations in early human milk but not in cord blood samples, and that, although prenatal PCB exposure seems to be most effective in affecting later development, an additional effect of postnatal exposure through breastfeeding is likely. Results from a study in Michigan of children born to mothers with varying degrees of consumption of fish from Lake Michigan showed a memory deficit at 7 months5 and at 4 years of age6 which was associated mainly with PCB concentrations in cord serum. Upon follow-up at 11 years, negative associations were still reported between a prenatal exposure index constructed from PCB concentrations in maternal serum, cord serum, and maternal milk, and deficit of intelligence quotient.14 By contrast with the Michigan study, in-utero PCB exposure in a North Carolina cohort was estimated from PCB concentrations in maternal milk reflecting maternal body burden during pregnancy. Neurodevelopmental delay was reported to be related to transplacental PCB exposure up to 24 months of age but not beyond.7 No cognitive or motor deficit was reported to be associated with prenatal PCB exposure at later ages, however.8 Our observations corroborate these early findings in stressing the developmental relevance of both background PCB concentrations and prenatal exposure. In terms of risk assessment these early studies have to be assessed relative to the decline of environmental PCB exposure that has taken place since about 1980. Correspondingly, PCB concentrations in human milk have also decreased over the years.18 However, the human infant by way of breastfeeding is exposed to PCB doses which, on a bodyweight basis, exceed those of adults by at least two orders of magnitude.24 It is, therefore, important to consider more recent observations on the developmental impact of PCBs at current background levels of exposure and to compare them with the outcome of the present study. In the Dutch Breast Milk Study9 at PCB concentrations similar to those of the present study (median milk PCB=405 ng/g fat; median cord PCB=0·38 ng/mL) associations were reported between PCBs/dioxins in maternal plasma and psychomotor development at 3 months of age, and with postnatal lactational exposure at 7 months in the breastfed children with the highest PCB concentrations. However, at 18 months of age these exposure-related effects were no longer seen.25 Upon follow-up of the Dutch cohort at 42 months, cognitive delay in terms of reduced Kaufman scores was recorded with increasing PCB concentrations in maternal plasma during pregnancy but not as clearly
1606
with PCBs in cord blood or early human milk.15 Our findings are largely in line with the Dutch findings, though at 42 months there are discrepancies in terms of the effective PCB matrix. In our study increasing PCB concentrations in early maternal milk predicted decreasing Kaufman scores, whereas in the Dutch study mainly PCB concentrations in maternal plasma samples were predictive for poorer Kaufman scores. No clearcut explanation for this difference can as yet be given. Analytical laboratory differences should be considered and, unlike in the Dutch studies, psychodevelopmental delay has also been related to PCBs in early maternal milk in some studies outside Europe.8,14 Furthermore, the lean matrix of cord blood with associated analytical difficulties could explain the lack of associations of cord PCB with later psychodevelopmental endpoints in our study. A particularly interesting finding of our study is the observation that, based on two different exposure-models, postnatal PCB dose via lactation exhibited negative association with scores on the Kaufman Assessment Battery for Children at 42 months but not with Bayley Scales of Infant Development scores at 30 months of age. It is tempting to try to explain this observation by comparing both diagnostic tools in terms of underlying neuropsychological requirements. The Bayley Scales of Infant Development is a developmental test based on the developmental milestone concept,26 whereas the Kaufman Assessment Battery for Children is based on a model of simultaneous and sequential information processing19 covering higher order cognitive operations relative to the Bayley Scales of Infant Development. It is a challenge for subsequent confirmation and more in-depth research to test the hypothesis, that higher-order mental processes as covered by the Kaufman Assessment Battery for Children develop later in early postnatal life and may, thus, be more prone to disruption by postnatal PCB exposure than the developmental milestones of the Bayley Scales of Infant Development. Due to the observational nature of this study the causative role of PCBs in producing developmental adversity cannot be considered proven. Since over 90% of PCB exposure in human beings is dietary, namely through animal fat,24 coexposure to other persistent aromatic compounds—eg, dioxins, furanes—is unavoidable;9 their neurotoxicity is less well documented, however. But, experimental data provide clear evidence for the developmental neurotoxicity of PCBs, document their likely prenatal nature,27 and suggest plausible biological mechanisms.4 As for the precise mechanisms of such action no definite answers can as yet be given. However, hypothyroid activity of PCBs or their metabolites has been proposed as an explanation.28 The regulating role of thyroid hormones in brain development is well established. In a clinical context dietary iodine deficiency results in a severe hypothyroidism known as cretinism associated with mental retardation. Congenital hypothyroidism is also associated with mental retardation, if untreated. PCBs and metabolites have been shown experimentally to interfere with thyroid hormone function through competitive interaction with thyroid hormones for binding to serum transport proteins and through induction of liver enzymes that metabolise thyroid hormones. Subclinical changes in thyroid hormones have also been associated with PCB exposure in several epidemiological studies. However, the link between PCBinduced thyroid dysfunction and PCB-related neurodevelopmental deficits still remains to be established.29
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
The positive effect quality of the home environment had on mental and motor development, despite the restricted range of the variable observed in our study, suggests that a favourable home environment might counteract the adverse developmental effects of PCBs.
11
12
Contributors G Winneke and H J Steingrüber with the help of E Schmidt developed the concept of the cohort study and worked on the initial grant proposal. E Schmidt helped to recruit the birth cohort and, together with H J Steingrüber and G Winneke, supervised the developmental fieldwork. J A Wiener, J Walkowiak, and S Wundram did the fieldwork and kept contact with the families. A Fastabend and B Heinzow were responsible for the analytical measurements. U Krämer supervised the epidemiological and statistical layout, together with J Walkowiak, who also constructed and maintained the database, and did the statistical work. J Walkowiak, U Krämer, and G Winneke worked on the manuscript and all authors read and commented on the paper.
Acknowledgments This study was partly supported within EU-programmes “Environment and Health” and “Environment and Climate” under contracts EV5VCT920 702 and EVN4-CT96-0209 by the European Commission (Brussels). We thank B Seidel and P Kues for their help in recruiting the cohort of newborns into the study. We acknowledge the careful analytical support of M Turfeld in measuring lead in cord blood. Helpful comments on an early version of the manuscript were given by H Roels (Brussels) and J Staessen (Leuven). Thanks are also due to the families for their continued long-term interest in the study and their willingness to participate over so many years.
13
14
15
16
17
18
19
References 1
Bradley RH. Children’s home environments, health, behavior and intervention efforts: a review using HOME inventory as a marker measure. Genet Soc Gen Psychol Monogr 1993; 119: 437–90. 2 Bradley RH, Caldwell BM. Home observation for measurement of the environment: administration Manual. Little Rock: University of Arkansas, 1984. 3 Riley EP, Vorhees CV eds. Handbook of behavioral teratology. New York and London: Plenum Press, 1986. 4 Seegal RF. Epidemiological and experimental evidence of PCBinduced neurotoxicity. Crit Rev Toxicol 1996; 26: 709–39. 5 Jacobson SW, Fein G, Jacobson, JL, Schwartz PM, Dowler, JK. The effect of intrauterine PCB exposure on visual recognition memory. Child Dev 1985; 56: 853–60. 6 Jacobson JL, Jacobson SW, Humphrey HEB. Effects of in utero exposure to polychlorinated biphenyls and related contaminants on cognitive functioning in young children. J Pediatr 1990; 116: 38–45. 7 Rogan WJ, Gladen, BC, McKinney JD, et al. Neonatal effects of transplacental exposure to PCBs and DDE. J Pediatr 1986; 109: 335–41. 8 Gladen BC, Rogan WJ, Hardy P, et al. Development after exposure to polychlorinated biphenyls and dichlorodiphenyl dichloroethene transplacentally and through human milk. J Pediatr 1988; 113: 991–95. 9 Koopmann-Esseboom C, Huisman M, Weisglas-Kuperus N, et al. PCB and dioxin levels in plasma and human milk of 418 Dutch women and their infants: predictive value of PCB congener level in maternal plasma for fetal and infant’s exposure to PCBs and dioxins. Chemosphere 1994; 28: 1721–32. 10 Huisman M, Koopman-Esseboom C, Fidler V, et al. Perinatal
THE LANCET • Vol 358 • November 10, 2001
20
21 22
23
24
25
26 27
28
29
exposure to polychlorinated biphenyls and dioxins and its effect on neonatal neurological development. Early Hum Dev 1995; 41: 111–27. Schantz SL. Developmental neurotoxicity of PCBs in humans: what do we know and where do we go from here? Neurotoxicol Teratol 1996; 18: 217–27. Winneke G, Bucholski A, Heinzow B, et al. Developmental neurotoxicity of polychlorinated biphenyls (PCBs): cognitive and psychomotor functions in 7-months-old children. Toxicol Lett 1998; 102: 423–28. Bellinger D. Interpreting the literature on lead and child development: the neglected role of the ”experimental system”. Neurotoxicol Teratol 1995; 17: 201–12. Jacobson JL, Jacobson SW. Intellectual impairment in children exposed to polychlorinated biphenyls in utero. N Engl J Med 1996; 335: 783–89. Patandin S, Lanting CI, Mulder PGH, Boersma ER, Sauer PJJ, Weisglas-Kuperus N. Effects of environmental exposure to polychlorinated biphenyls and dioxins on cognitive abilities in Dutch children at 42 months of age. J Pediatr 1999; 134: 33–41. Walkowiak J. Häusliche Umwelt und frühkindliche Entwicklung. Ein Beitrag zur Etablierung der HOME-Skala in Deutschland. PhDThesis. University of Düsseldorf; 2000. Fastabend A. Bestimmung umweltbedingter Konzentrationen von PCBs und Organochlor-pestiziden in Blut mit GC-ECD unter Anwendung optimierter Probenvorbereitungsverfahren. PhD-Thesis. University of Clausthal; 2000. Schade G, Heinzow B. Organochlorine pesticides and polychlorinated biphenyls in human milk of mothers living in Northern Germany: current extent of contamination, time trend from 1986 to 1997, and factors that influence the level of contamination. Sci Total Environ 1998; 215: 31–39. Melchers P, Preus U. Kaufman Assessment Battery for Children (K-ABC). Deutschsprachige Fassung, 2. Aufl. Lisse (NL): Swets and Zeitlinger; 1994. Ganzeboom BG, Treiman DJ. Internationally comparable measures of occupational status for the 1988 International Standard Classification of Occupations. Soc Sci Res 1996; 25: 201–39. Tewes U. Hamburg-Wechsler-Intelligenztest für Erwachsene – Revision 1991 (HAWIE-R).- Bern: Verlag Hans Huber; 1991. Elardo R, Bradley R, Caldwell B. The relation of infant’s home environments to mental test performance from 6 to 36 months: a longitudinal analysis. Child Dev 1975; 46: 71–76. Hansen LG. Stepping backward to improve assessment of PCB congener toxicities. Environ Health Perspect 1998; 106 (suppl 1): 171–89. Umweltbundesamt. Stoffmonographie PCB – Referenzwerte für Blut. Bundesgesundheitsblatt – Gesundheitsforschung – Gesundheitsschutz,1999; 42: 511–21. Koopmann-Esseboom C, Weisglas-Kuperus N, de Ridder MAJ, van der Paauw CG, Tuinstra LGMT, Sauer PJJ. Effects of polychlorinated biphenyl/dioxin exposure and feeding type on infant’s mental and psychomotor development. Pediatrics 1996; 97: 700–06. Bayley N. Bayley Scales of Infant Development (2nd edn). San Antonio, TX: Psychological Corporation; 1993. Lilienthal H, Winneke G. Sensitive periods for behavioral toxicity of polychlorinated biphenyls: determination by cross-fostering in rats. Fundam Appl Toxicol 1991, 17: 368–75. Porterfield S. Vulnerability of the developing brain to thyroid abnormalities: environmental insults to the thyroid system. Environ Health Perspect 1994; 102: 125–30. Brouwer A, Longnecker MP, Birnbaum LS, et al. Characterization of potential endocrine-related health effects of low-dose levels of exposure to PCBs. Environ Health Perspect 1999; 107: 639–49.
1607
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Environmental exposure to polychlorinated biphenyls and quality of the home environment: effects on psychodevelopment in early childhood Jens Walkowiak, Jörg-A Wiener, Annemarie Fastabend, Birger Heinzow, Ursula Krämer, Eberhard Schmidt, Hans-J Steingrüber, Sabine Wundram, Gerhard Winneke
Summary Background There is uncertainty whether environmental levels of exposure to polychlorinated biphenyls (PCBs) adversely affect mental and motor development in early childhood. We aimed to establish whether such an effect is of only prenatal or additional postnatal origin, and if a favourable home environment can counteract this effect. Methods Between 1993 and 1995 we recruited 171 healthy mother-infant pairs and prospectively measured psychodevelopment in newborn infants aged 7, 18, 30, and 42 months. We estimated prenatal and perinatal PCB exposure of newborn babies in cord blood and maternal milk. At 42 months we measured postnatal PCB concentrations in serum. At 18 months the quality of the home environment was assessed using the Home Observation for Measurement of the Environment scale. Mental and psychomotor development of the children were assessed using the Bayley Scales of Infant Development until 30 months and the Kaufman Assessment Battery for Children at 42 months. Findings Negative associations between milk PCB and mental/motor development were reported at all ages, becoming significant from 30 months onwards. Over 30 months, for a PCB increase from 173 (5th percentile) to 679 ng/g lipids in milk (95th percentile) there was a decrease of 8·3 points (95% CI –16·5 to 0·0) in the Bayley Scales of Infant Development mental scores, and a 9·1 point decrease (95% CI –17·2 to –1·02) in the Bayley Scales of Infant Development motor scores. There was also a negative effect of postnatal PCB exposure via breastfeeding at 42 months. Home environment had a positive effect from 30 months onwards (Bayley Scales of Infant Development mental score increase of 9·4 points [95% CI 2·2–16·7]). Interpretation Prenatal PCB exposure at current European background levels inhibits, and a favourable home environment supports, mental and motor development until 42 months of age. PCB exposure also has an effect postnatally. Lancet 2001; 358: 1602–07 See Commentary page 1568 Divisions of Analytical Chemistry, Epidemiology, and Neurobehavioral Toxicology at Medical Institute of Environmental Hygiene (A Fastabend PhD, U Krämer PhD, J Walkowiak PhD, Prof G Winneke PhD, S Wundram), and Institute of Medical Psychology (J A Wiener, Prof H J Steingrüber PhD), and University Child Clinic, Medical Faculty (Prof E Schmidt MD), Heinrich-HeineUniversity Düsseldorf; and Laboratory of Environmental Toxicology at the State Agency of Nature and Environment of SchleswigHolstein, Flintbeck, Germany (B Heinzow MD) Correspondence to: Dr Gerhard Winneke, Medical Institute of Environmental Hygiene at Heinrich-Heine-Universität Düsseldorf, Auf’m Hennekamp 50, D-40225 Düsseldorf, Germany (e-mail: [email protected])
1602
Introduction Both endogenous and exogenous factors and their interactions regulate human child development. Among the endogenous factors, genetic and biochemical induction and modulation are predominant, whereas psychosocial and physicochemical factors and their interactions are important exogenous determinants of general development or psychodevelopment of the human infant. As for the psychosocial dimension many different individual aspects of the child environment have received attention in developmental studies.1 However, the isolated study of such variables underestimates the contribution of the many interacting environmental components for development. Therefore, more comprehensive constructs for a quantitative description of relevant features of home environments have been developed. A prominent instrument in this respect is the Home Observation for Measurement of the Environment (HOME).2 We have used the score to compare the developmental effects of the quality of the home environment with those of early chemical exposure. Among the exogenous factors affecting early neurobehavioural child development is the exposure to chemicals in utero or during the early stages of brain development.3 Some environmental chemicals, such as inorganic lead or organic mercury, have received attention in this respect, including the polychlorinated biphenyls (PCBs). PCBs are persistent environmental contaminants consisting of up to 209 individual congeners. PCBs and their metabolites cross the placenta, exposing the vulnerable fetus to PCBs circulating in maternal blood. After birth the infant is additionally exposed to relatively high PCB concentrations in human milk. Among a broad spectrum of biological effects, developmental neurotoxicity seems to be a prominent feature of these chemical mixtures, as can be judged from findings in animals and human beings.4 Evidence of neurodevelopmental effects of prenatal or early postnatal PCB exposure in human beings at environmental exposure levels is largely based on findings from four sets of cohort studies.5–10 Although all of these studies did report some adverse effects of early PCB exposure on neurological or cognitive development, they are not fully consistent with regard to confounding, spectrum, persistence of effects, or the biological matrix in which PCBs were reported most predictive for later developmental effects.11 Previous reports describing neurodevelopmental effects of environmental chemicals have typically concentrated on exposure-response effects in isolation, and have treated other factors of developmental impact merely as confounders. We have compared the independent effects of early PCB-exposure and the home environment from the age of 7–42 months. Part of the findings at 7 months has already been reported;12 they are covered again in a modified manner here, to allow for direct comparisons with the observations made at the later ages.
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Methods Study population We recruited 171 healthy mother-infant pairs between October, 1993, and May, 1995, from the obstetrical wards of three Düsseldorf hospitals. Three advanced medical students were involved in the recruitment process. Criteria for inclusion were: principal agreement of the mother, first and second born babies delivered at term (weeks 37–42 of pregnancy) from native German families; an Apgar score at 5 minutes of at least 78; and no serious illnesses or complications during pregnancy and delivery. The restriction to parities one and two served to keep the sample more homogeneous with respect to PCB exposure. Those fulfilling these criteria and having an appropriate medical condition were approached for participation, and about 70% of these, with written consent, agreed to participate until their babies were at least 7 months of age. The study protocol was approved by the medical ethics committee of the Heinrich-HeineUniversity, Düsseldorf, Medical Faculty. Quality of the home environment The HOME score by Caldwell and Bradley2 is a powerful and validated instrument for the assessment of the home environment both as a developmental determinant and an influential confounder in neurotoxicity studies.13–15 The infant version (0–3 years) used here consists of 45 items covering emotional and verbal responsivity of the mother, acceptance of the child’s behaviour, organisation of the home environment, availability of toys, parental involvement with the child, and variability of daily experience. Assessment takes place during a semistructured interview with the mother at home in the presence of the child. We used a version adapted to fit the German cultural context16 at 18 months of age. Retest reliabilities, checked by two observers in an independent sample of 26 children at ages 5 and 16 months (observer A) as well as ages 7 and 18 months (observer B), were rtt=0·71–0·72 for these two 11 month intervals.16 PCB exposure We used the sum of the three PCB congeners 138, 153, and 180 as the marker of PCB exposure. We took cord blood samples in polyethylene monovettes and stored them at –20˚C until analysis according to Fastabend.17 At 42 months we drew venous blood (5 mL) using vacutainer glass tubes for the separation of serum. Blood samples were allowed to clot and then centrifuged at 3000 rpm. We transferred the serum into glass vials and stored them at –20˚C until sample preparation and analysis. For PCB analyses in human milk spot samples (5–10 mL) were collected around 2 weeks postpartum in glass vials, and stored at –20˚C until analysis according to a standard method.18 Briefly, after thawing and homogenisation, we extracted the PCBs by n-heptane and purified the extracts by a silica gel column. We did PCB analyses on a highresolution gas chromatograph with electron capture detection. Three marker congeners 138, 153, and 180 were determined with a detection limit of 0·01 µg/L (serum) and 0·005 mg/kg fat (milk). We determined the lipid content with the photometric Mercko-Test 3221 (Merck, Darmstadt). Psychodevelopmental tests Psychomotor and cognitive development were assessed at 7, 18, 30, and 42 months. Testing was done at the child’s home, usually in the presence of the mother, by the same examiner (JAW), who was unaware of the child’s PCB
THE LANCET • Vol 358 • November 10, 2001
concentration, for the whole study period within 2 weeks of the target age. Of the Bayley scales of Infant Development, an established psychodevelopmental tool, the mental and the motor scales were used. The mental scale assesses the child’s level of cognitive functioning, language development, and personal/social development. The motor scale assesses fine, and gross motor functioning. For reasons of comparability raw scores were transformed to yield a mean of 100 and an SD of 15. At 42 months of age, we measured intelligence with the German version of the Kaufman Assessment Battery for Children.19 Scores from the Sequential Processing and Simultaneous Processing subscales are combined to yield the Mental Processing Composite-Index which is standardised to a mean of 100 and an SD of 15. Potential confounders/covariates Potential confounders included age at examination (days), gestational age (weeks), alcohol/smoking during pregnancy (separately scored as 0/1), Apgar score (8, 9, 10), neonatal illness/jaundice (0/1), spontaneous delivery (0/1), parity (1/2), lead in cord blood (µg/L), chronic diseases during past year (0/1), duration of breastfeeding (months), parental education (years of schooling; maximum of either parent), parental occupation (10 categories;20 highest of either parent), body-mass index of mother (kg/m2), and maternal verbal intelligence quotient (subtest vocabulary from the German form of the Wechsler Adult Intelligence Scale).21 Statistical analysis We treated the data both descriptively as well as inferentially with SAS (version 8.01). To analyse the association of HOME and PCBs with the target variables we used multiple linear regression. Model building involved two steps: in the first step, variables were selected based on theoretical knowledge of their effect on mental or motor development, or on empirical evidence from previous studies on the relation of either PCBs or the HOME to the respective outcome variables. In the second step, selection of variables was data-driven; only those variables were selected which had correlations at p<0·20 with both the exposure index and at least two of the outcome variables after inclusion of the first step variables. The variables in the first step were parental education, sex, maternal intelligence quotient, as well as HOME and PCB simultaneously. Those in the second step were parity, smoking in pregnancy, and body-mass index. All outcome variables were uniformly adjusted for these variables. Interactions were considered but discarded if not significant. Additionally, to take advantage of the time-series aspect of the Bayley Scales of Infant Development administered at three points in time, a repeated measurements analysis was also done. We used one-tailed probabilities throughout, because the hypotheses were directional, and effects at p<0·05 were judged significant.
Results From recruitment until 42 months of age the sample size dropped from an initial 171 to 116. 126 mothers from the initial 171 provided milk samples at 2 weeks and 91 of these remained in the study until final testing at 42 months. Despite this attrition the sample structure changed only marginally. No significant difference from recruitment to testing at 42 months was detected for PCB concentrations, sex, birthweight, gestational/maternal age, neonatal condition, parental education, maternal
1603
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
PCB concentrations
Percentiles n*
Cord PCB (ng/mL) 141 Milk PCB (ng/g lipids) 126 Serum PCB at 42 92 months (ng/mL)
5%
median 75%
95%
0·11 0·28 0·39 0·50 0·83 173 294 404 535 679 0·23 0·49 1·22 1·87 3·35
Duration of breastfeeding (PCB ⭐2 weeks (ng/mL) 15 >2 weeks to 4 months 32 (ng/mL) >4 months (ng/mL) 44 HOME (total score)† Sum of six subscales 156
25%
42 months)* 0·14 0·24 0·31 0·49
0·36 0·68
0·52 1·20
4·39 1·79
0·29
1·77
2·50
3·52
34
1·37 38
41
42
44
Number of breastfed children is 91, one child with missing data on duration of nursing. *Different sample sizes due to incomplete information. †Raw-scores of the HOME scale administered at 18 months. The theoretical maximum is 45.
Table 1: Cord, milk and serum PCB concentrations at different ages and relative to breastfeeding, and HOME scores
intelligence quotient, neonatal illness, alcohol/smoking in pregnancy, and lead in cord blood. However, there was a tendency for families staying within the study to be better educated. Table 1 shows that PCB concentrations in serum samples at 42 months increased markedly with duration of breastfeeding and were about five times higher than those in the group of non-breastfed children. Children with a breastfeeding period of 2 weeks or less were included in the group of non-breastfed children. Because the PCB distributions were skewed all subsequent analyses are based on log2-transformed values. The correlations between PCB in cord blood and milk were positive and highly significant (r=0·57; p<0·0001). The range of HOME scores was quite narrow and were shifted to the upper end of the environmental quality scale (table 1). Mean maternal age (SD) was 29·5 years (4·4), mean gestational age was 39·9 weeks (1·2), and the average birthweight was 3493 g (463). Lead concentrations in cord blood were very low (20·3 [7·1] µg/L). Mean raw scores of the Bayley Scales of Infant Development mental scale increased from 64·5 (SD 2·2) at 7 months to 149·3 (5·1) at age 30 months. For the Bayley Scales of Infant Development motor scale this increase was from 40·7 (3·5) at 7 months to 94·8 (2·9) at 30 months. For the Kaufman Assessment Battery for Children at 42 months the mean score was 103·5 (10·9). Table 2 shows results from the regression analyses relating PCB concentrations in maternal milk and the HOME scale to mental and motor development. After adjustment, exposure-response associations with PCB were negative and of only borderline significance at ages 7 and 18 months, but became significant at 30 (Bayley Scales of Infant Development) and at 42 months of age (Kaufman Assessment Battery for Children). Treating the age-range of 7–30 months as a time-series, early PCB exposure shows significant negative overall associations
with mental/motor development. Only children with complete data were included. However, the strength of associations did not change when using simple imputation techniques to account for missing values. Negative associations with PCB increased with age, and positive associations with the quality of the home environment became more pronounced at 30 or 42 months of age. Dose-response associations at 30 and 42 months of age, based on adjusted values for the Bayley Scales of Infant Development mental/motor development and the Mental Processing Composite-Index of the Kaufman Assessment Battery for Children, reveal a pattern of facilitation for the home environment and one of inhibition for milk PCB with rather similar slopes for both variables over the range of observed HOME scores or milk PCB concentrations (figure 1). Yet, the overall positive impact of the home environment on development is stronger than the negative effect of neonatal PCB exposure. This is evident from figure 2 which illustrates effectsizes derived from the respective regression coefficients from table 2. Here, the changes of the Bayley Scales of Infant Development scores or Kaufman Assessment Battery for Children scores are given for an increase of HOME scores or PCB concentrations from the lower 5th to the upper 95th of their distributions from table 1. To exemplify for the 30-months assessment of Bayley Scales of Infant Development mental score, the increase with increasing HOME score was 17·7, whereas the decrease with increasing milk PCB concentrations was just 9·9 points. However, for the time-series approach covering the full 7–30 months period (not given in figure 2) effect-size differences between HOME and milk PCB concentrations were less pronounced: the overall increase for Bayley Scales of Infant Development (mental) with increasing HOME score was 9·4 points (95% CI 2·2–16·4), and the overall decrease with increasing PCB concentraion was –8·3 points (–16·5 to 0·0); a very similar pattern emerged for Bayley Scales of Infant Development motor development: an increase with increasing HOME of 9·8 points (2·7–16·9) and a decrease with increasing PCB concentrations of –9·1 (–17·2 to –1·02). All associations with cord blood PCB and the Bayley Scales of Infant Development mental score were small and even slightly positive. Association between the Bayley Scales of Infant Development motor score and PCB was negative and very small. For example, for the time-series model and using the uniform confounder model the final outcome was as follows: t1,98=–0·93, p=0·36. The associations between cord blood PCB and the Mental Processing Composite-Index of the Kaufman Assessment Battery for Children at 42 months were small and even slightly positive. The possibility of an additional developmental impact of postnatal PCB intake which, at this age, occurs mainly through breastfeeding (table 1), was tested by means of two models. The first model used a rough estimate of
Mental development PCBmilk log2 [ng/g]
7 months (BSID; n=110) 18 months (BSID; n=112) 30 months (BSID; n=104) 7–30 months (cohort analysis; n=104) 42 months (K-ABC; n=87)
Motor development HOME score
PCBmilk log2 [ng/g]
HOME score
*
t
p†
*
t
p†
*
t
p†
*
t
p†
–3·61 –4·11 –4·98 –4·19
–1·26 –1·56 –1·80 –1·99
0·10 0·06 0·035 0·025
0·56 0·65 1·77 0·94
1·14 1·47 3·74 2·58
0·13 0·07 0·0002 0·005
–3·13 –4·78 –4·73 –4·61
–1·19 –1·71 –1·68 –2·22
0·12 0·045 0·05 0·015
0·47 0·88 1·82 0·98
1·04 1·88 3·77 2·70
0·15 0·03 0·0002 0·004
–4·30
–1·93
0·028
1·29
3·35
0·0006
··
··
··
··
··
··
K-ABC=Kaufman Achievement Battery for Children. BSID=Bayley Scales of Infant Development. *Regression coefficients. † One-tailed probabilities.
Table 2: Association of PCR and HOME with mental and motor development
1604
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Bayley (standardised values)
Mental development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <38
38–39
40–41
42
>42
Bayley (standardised values)
ARTICLES
Mental development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <276
Bayley (standardised values)
Motor development (Bayley) measured at 30 months 114 108 102 96 90 84 78 <38
38–39
40–41
42
>42
Motor development (Bayley) measured at 30 months
<276
42
>42
Bayley (standardised values)
Kaufman (standardised values)
40–41
>276–352 >352–448 >448–558
>558
Milk PCBs (ng/g lipids)
Mental development (Kaufman) measured at 42 months 114 108 102 96 90 84 78 38–39
>558
114 108 102 96 90 84 78
HOME scale (raw scores)
<38
>276–352 >352–448 >448–558
Milk PCBs (ng/g lipids) Bayley (standardised values)
HOME scale (raw scores)
Mental development (Kaufman) measured at 42 months 114 108 102 96 90 84 78 <276
>276–352 >352–448 >448–558
>558
Milk PCBs (ng/g lipids)
HOME scale (raw scores)
Figure 1: Exposure-response associations for HOME scores or milk PCB and mental or psychomotor development Dose-response associations for Bayley Scales of Infant Development mental and motor scores and Kaufman Assessment Battery for Children scores in relation to quality of the home environment and PCB concentrations (quintiles). Values are means (95% CI) and regression lines are superimposed.
Mental HOME PCBs (milk)
Psychomotor HOME PCBs (milk) K–ABC
BSID
20
‡
‡
Change in score
15
‡
10
† *
5
dose, namely (milk PCB⫻months of breastfeeding) adjusted for prenatal or perinatal exposure. This model was applied to the repeated measurement approach of Bayley Scales of Infant Development (mental and motor) and to the Kaufman Assessment Battery for Children (Mental Processing Composite-Index). The second model was applicable to the Kaufman Assessment Battery for Children as the outcome variable only, and used PCB in 42-months serum, again adjusted for prenatal or perinatal exposure. No significant or borderline association with calculated postnatal PCB exposure was reported for both Bayley Scales of Infant Development measures. However, a significant negative effect of postnatal exposure was noted for both models as applied to the Kaufman Assessment Battery for Children at 42 months of age (t=–1·89, p=0·031 for dose, and t=–2·01, p=0·025 for PCB at 42 months).
0
Discussion –5 *
–10 7 months
†
† †
18 months
30 months
†
42 months
Figure 2: Adjusted effect sizes for mental and psychomotor development in relation to the quality of the home environment or milk PCB concentrations Calculated from the regression coefficients given in table 2 for a change in the independent variables from the 5th–95th percentiles (table 1).*p<0·10; †p<0·05; ‡p<0·001 (one-tailed). Upward columns represent HOME-related developmental facilitation, downward columns represent PCB-related developmental delay or deficit.
THE LANCET • Vol 358 • November 10, 2001
Despite considerable attrition of the cohort during the study period, mainly due to loss of interest, moving away from Düsseldorf, or some cases of initially unidentified congenital disorders, we observed a strong positive effect of the home environment as assessed by the HOME scale for psychodevelopment. With increasing age the degree of association between HOME and mental or motor development increased, an observation which is consistent with published evidence.22 This effect might reflect the extended effect of the quality of the child’s environment, but could also be due to the increase with age of the psychometric quality of developmental testing. Although
1605
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
the range of the HOME scores was small and restricted to the upper end of the continuum, the size of the facilitating effect of the home environment on psychodevelopment was remarkable. Despite their ban in terms of production and use between 1977 and 1983, PCBs are still of environmental concern. The 209 possible PCB congeners differ by the number and position of chlorine atoms on the two benzene rings making up the basic structure of the molecule. From the six marker congeners (28, 52, 101, 138, 153, 180), typically measured to characterise PCB concentrations in environmental media, we only used the three higher chlorinated compounds as indicators of the internal PCB exposure. This is in accordance with recommendations23 and also considers the fact that these three account for about 60% of the total PCB content in human tissue samples.24 Having defined prenatal and neonatal PCB exposure in this manner we have shown that mental and motor development between 7 and 42 months of age has a significant negative association with PCB concentrations in early human milk but not in cord blood samples, and that, although prenatal PCB exposure seems to be most effective in affecting later development, an additional effect of postnatal exposure through breastfeeding is likely. Results from a study in Michigan of children born to mothers with varying degrees of consumption of fish from Lake Michigan showed a memory deficit at 7 months5 and at 4 years of age6 which was associated mainly with PCB concentrations in cord serum. Upon follow-up at 11 years, negative associations were still reported between a prenatal exposure index constructed from PCB concentrations in maternal serum, cord serum, and maternal milk, and deficit of intelligence quotient.14 By contrast with the Michigan study, in-utero PCB exposure in a North Carolina cohort was estimated from PCB concentrations in maternal milk reflecting maternal body burden during pregnancy. Neurodevelopmental delay was reported to be related to transplacental PCB exposure up to 24 months of age but not beyond.7 No cognitive or motor deficit was reported to be associated with prenatal PCB exposure at later ages, however.8 Our observations corroborate these early findings in stressing the developmental relevance of both background PCB concentrations and prenatal exposure. In terms of risk assessment these early studies have to be assessed relative to the decline of environmental PCB exposure that has taken place since about 1980. Correspondingly, PCB concentrations in human milk have also decreased over the years.18 However, the human infant by way of breastfeeding is exposed to PCB doses which, on a bodyweight basis, exceed those of adults by at least two orders of magnitude.24 It is, therefore, important to consider more recent observations on the developmental impact of PCBs at current background levels of exposure and to compare them with the outcome of the present study. In the Dutch Breast Milk Study9 at PCB concentrations similar to those of the present study (median milk PCB=405 ng/g fat; median cord PCB=0·38 ng/mL) associations were reported between PCBs/dioxins in maternal plasma and psychomotor development at 3 months of age, and with postnatal lactational exposure at 7 months in the breastfed children with the highest PCB concentrations. However, at 18 months of age these exposure-related effects were no longer seen.25 Upon follow-up of the Dutch cohort at 42 months, cognitive delay in terms of reduced Kaufman scores was recorded with increasing PCB concentrations in maternal plasma during pregnancy but not as clearly
1606
with PCBs in cord blood or early human milk.15 Our findings are largely in line with the Dutch findings, though at 42 months there are discrepancies in terms of the effective PCB matrix. In our study increasing PCB concentrations in early maternal milk predicted decreasing Kaufman scores, whereas in the Dutch study mainly PCB concentrations in maternal plasma samples were predictive for poorer Kaufman scores. No clearcut explanation for this difference can as yet be given. Analytical laboratory differences should be considered and, unlike in the Dutch studies, psychodevelopmental delay has also been related to PCBs in early maternal milk in some studies outside Europe.8,14 Furthermore, the lean matrix of cord blood with associated analytical difficulties could explain the lack of associations of cord PCB with later psychodevelopmental endpoints in our study. A particularly interesting finding of our study is the observation that, based on two different exposure-models, postnatal PCB dose via lactation exhibited negative association with scores on the Kaufman Assessment Battery for Children at 42 months but not with Bayley Scales of Infant Development scores at 30 months of age. It is tempting to try to explain this observation by comparing both diagnostic tools in terms of underlying neuropsychological requirements. The Bayley Scales of Infant Development is a developmental test based on the developmental milestone concept,26 whereas the Kaufman Assessment Battery for Children is based on a model of simultaneous and sequential information processing19 covering higher order cognitive operations relative to the Bayley Scales of Infant Development. It is a challenge for subsequent confirmation and more in-depth research to test the hypothesis, that higher-order mental processes as covered by the Kaufman Assessment Battery for Children develop later in early postnatal life and may, thus, be more prone to disruption by postnatal PCB exposure than the developmental milestones of the Bayley Scales of Infant Development. Due to the observational nature of this study the causative role of PCBs in producing developmental adversity cannot be considered proven. Since over 90% of PCB exposure in human beings is dietary, namely through animal fat,24 coexposure to other persistent aromatic compounds—eg, dioxins, furanes—is unavoidable;9 their neurotoxicity is less well documented, however. But, experimental data provide clear evidence for the developmental neurotoxicity of PCBs, document their likely prenatal nature,27 and suggest plausible biological mechanisms.4 As for the precise mechanisms of such action no definite answers can as yet be given. However, hypothyroid activity of PCBs or their metabolites has been proposed as an explanation.28 The regulating role of thyroid hormones in brain development is well established. In a clinical context dietary iodine deficiency results in a severe hypothyroidism known as cretinism associated with mental retardation. Congenital hypothyroidism is also associated with mental retardation, if untreated. PCBs and metabolites have been shown experimentally to interfere with thyroid hormone function through competitive interaction with thyroid hormones for binding to serum transport proteins and through induction of liver enzymes that metabolise thyroid hormones. Subclinical changes in thyroid hormones have also been associated with PCB exposure in several epidemiological studies. However, the link between PCBinduced thyroid dysfunction and PCB-related neurodevelopmental deficits still remains to be established.29
THE LANCET • Vol 358 • November 10, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
The positive effect quality of the home environment had on mental and motor development, despite the restricted range of the variable observed in our study, suggests that a favourable home environment might counteract the adverse developmental effects of PCBs.
11
12
Contributors G Winneke and H J Steingrüber with the help of E Schmidt developed the concept of the cohort study and worked on the initial grant proposal. E Schmidt helped to recruit the birth cohort and, together with H J Steingrüber and G Winneke, supervised the developmental fieldwork. J A Wiener, J Walkowiak, and S Wundram did the fieldwork and kept contact with the families. A Fastabend and B Heinzow were responsible for the analytical measurements. U Krämer supervised the epidemiological and statistical layout, together with J Walkowiak, who also constructed and maintained the database, and did the statistical work. J Walkowiak, U Krämer, and G Winneke worked on the manuscript and all authors read and commented on the paper.
Acknowledgments This study was partly supported within EU-programmes “Environment and Health” and “Environment and Climate” under contracts EV5VCT920 702 and EVN4-CT96-0209 by the European Commission (Brussels). We thank B Seidel and P Kues for their help in recruiting the cohort of newborns into the study. We acknowledge the careful analytical support of M Turfeld in measuring lead in cord blood. Helpful comments on an early version of the manuscript were given by H Roels (Brussels) and J Staessen (Leuven). Thanks are also due to the families for their continued long-term interest in the study and their willingness to participate over so many years.
13
14
15
16
17
18
19
References 1
Bradley RH. Children’s home environments, health, behavior and intervention efforts: a review using HOME inventory as a marker measure. Genet Soc Gen Psychol Monogr 1993; 119: 437–90. 2 Bradley RH, Caldwell BM. Home observation for measurement of the environment: administration Manual. Little Rock: University of Arkansas, 1984. 3 Riley EP, Vorhees CV eds. Handbook of behavioral teratology. New York and London: Plenum Press, 1986. 4 Seegal RF. Epidemiological and experimental evidence of PCBinduced neurotoxicity. Crit Rev Toxicol 1996; 26: 709–39. 5 Jacobson SW, Fein G, Jacobson, JL, Schwartz PM, Dowler, JK. The effect of intrauterine PCB exposure on visual recognition memory. Child Dev 1985; 56: 853–60. 6 Jacobson JL, Jacobson SW, Humphrey HEB. Effects of in utero exposure to polychlorinated biphenyls and related contaminants on cognitive functioning in young children. J Pediatr 1990; 116: 38–45. 7 Rogan WJ, Gladen, BC, McKinney JD, et al. Neonatal effects of transplacental exposure to PCBs and DDE. J Pediatr 1986; 109: 335–41. 8 Gladen BC, Rogan WJ, Hardy P, et al. Development after exposure to polychlorinated biphenyls and dichlorodiphenyl dichloroethene transplacentally and through human milk. J Pediatr 1988; 113: 991–95. 9 Koopmann-Esseboom C, Huisman M, Weisglas-Kuperus N, et al. PCB and dioxin levels in plasma and human milk of 418 Dutch women and their infants: predictive value of PCB congener level in maternal plasma for fetal and infant’s exposure to PCBs and dioxins. Chemosphere 1994; 28: 1721–32. 10 Huisman M, Koopman-Esseboom C, Fidler V, et al. Perinatal
THE LANCET • Vol 358 • November 10, 2001
20
21 22
23
24
25
26 27
28
29
exposure to polychlorinated biphenyls and dioxins and its effect on neonatal neurological development. Early Hum Dev 1995; 41: 111–27. Schantz SL. Developmental neurotoxicity of PCBs in humans: what do we know and where do we go from here? Neurotoxicol Teratol 1996; 18: 217–27. Winneke G, Bucholski A, Heinzow B, et al. Developmental neurotoxicity of polychlorinated biphenyls (PCBs): cognitive and psychomotor functions in 7-months-old children. Toxicol Lett 1998; 102: 423–28. Bellinger D. Interpreting the literature on lead and child development: the neglected role of the ”experimental system”. Neurotoxicol Teratol 1995; 17: 201–12. Jacobson JL, Jacobson SW. Intellectual impairment in children exposed to polychlorinated biphenyls in utero. N Engl J Med 1996; 335: 783–89. Patandin S, Lanting CI, Mulder PGH, Boersma ER, Sauer PJJ, Weisglas-Kuperus N. Effects of environmental exposure to polychlorinated biphenyls and dioxins on cognitive abilities in Dutch children at 42 months of age. J Pediatr 1999; 134: 33–41. Walkowiak J. Häusliche Umwelt und frühkindliche Entwicklung. Ein Beitrag zur Etablierung der HOME-Skala in Deutschland. PhDThesis. University of Düsseldorf; 2000. Fastabend A. Bestimmung umweltbedingter Konzentrationen von PCBs und Organochlor-pestiziden in Blut mit GC-ECD unter Anwendung optimierter Probenvorbereitungsverfahren. PhD-Thesis. University of Clausthal; 2000. Schade G, Heinzow B. Organochlorine pesticides and polychlorinated biphenyls in human milk of mothers living in Northern Germany: current extent of contamination, time trend from 1986 to 1997, and factors that influence the level of contamination. Sci Total Environ 1998; 215: 31–39. Melchers P, Preus U. Kaufman Assessment Battery for Children (K-ABC). Deutschsprachige Fassung, 2. Aufl. Lisse (NL): Swets and Zeitlinger; 1994. Ganzeboom BG, Treiman DJ. Internationally comparable measures of occupational status for the 1988 International Standard Classification of Occupations. Soc Sci Res 1996; 25: 201–39. Tewes U. Hamburg-Wechsler-Intelligenztest für Erwachsene – Revision 1991 (HAWIE-R).- Bern: Verlag Hans Huber; 1991. Elardo R, Bradley R, Caldwell B. The relation of infant’s home environments to mental test performance from 6 to 36 months: a longitudinal analysis. Child Dev 1975; 46: 71–76. Hansen LG. Stepping backward to improve assessment of PCB congener toxicities. Environ Health Perspect 1998; 106 (suppl 1): 171–89. Umweltbundesamt. Stoffmonographie PCB – Referenzwerte für Blut. Bundesgesundheitsblatt – Gesundheitsforschung – Gesundheitsschutz,1999; 42: 511–21. Koopmann-Esseboom C, Weisglas-Kuperus N, de Ridder MAJ, van der Paauw CG, Tuinstra LGMT, Sauer PJJ. Effects of polychlorinated biphenyl/dioxin exposure and feeding type on infant’s mental and psychomotor development. Pediatrics 1996; 97: 700–06. Bayley N. Bayley Scales of Infant Development (2nd edn). San Antonio, TX: Psychological Corporation; 1993. Lilienthal H, Winneke G. Sensitive periods for behavioral toxicity of polychlorinated biphenyls: determination by cross-fostering in rats. Fundam Appl Toxicol 1991, 17: 368–75. Porterfield S. Vulnerability of the developing brain to thyroid abnormalities: environmental insults to the thyroid system. Environ Health Perspect 1994; 102: 125–30. Brouwer A, Longnecker MP, Birnbaum LS, et al. Characterization of potential endocrine-related health effects of low-dose levels of exposure to PCBs. Environ Health Perspect 1999; 107: 639–49.
1607
For personal use. Only reproduce with permission from The Lancet Publishing Group.