Background
Since polychlorinated biphenyls (PCBs) and methylmercury that are environmentally persistent toxicants cross the placenta, they may affect child development. Prenatal lead exposure seems to result in low birth weight [
1‐
6]. However, the effects of PCBs and methylmercury on fetal growth remain controversial [
7]: Some studies, performed in developed countries, reported adverse effects of prenatal PCB exposure on birth weight [
8‐
11]; by contrast, one study showed a positive association between them [
12]. Others failed to find such a significant association [
13‐
17]. Likewise, Foldspang and Hansen reported that high maternal and offspring methylmercury concentrations were associated with low birth weight [
18], but other researchers did not present a significant association between the exposure and outcome [
5,
19‐
21]. Apart from prenatal exposures via fish and seafood intake, the children of heavy and moderate smokers during pregnancy have a higher rate of being small for gestational age (SGA) than those of non-smokers [
22]. If the number of cigarettes smoked and passive smoking during pregnancy were not examined in a study [
8], absence of the principal cause would confuse a result derived from such toxic substances. Low birth weight is known to be associated with several chronic diseases in adults, including diabetes mellitus and hypertension [
23]. For the prevention of adult diseases, therefore, it is crucial to confirm which pollutants affect fetal growth.
In recent years, the exposure levels of environmental pollutants such as PCBs and lead are low in developed countries [
24,
25], predicting that the range of exposure (i.e., difference between the minimum and maximum) may not be wide enough to detect a statistically significant dose–response relationship. In this case, use of concurrent exposure model will be desirable for considering the interactive effect of potential substances other than the concerned one, even though the exposure levels were not so high. In fact, essential nutrients such as
n-3 polyunsaturated fatty acids (PUFAs) masked the effects of methylmercury [
26,
27]. Also, two previous studies demonstrated the association between maternal PCB concentrations and low birth weight only in male newborns [
28,
29], hypothesizing that male fetuses may be more vulnerable to such toxicants than female ones. Thus, careful selection of study subjects, exposure markers and possible confounders is sought when examining the complicated link between prenatal low-level exposures and fetal growth.
We have been performing a prospective birth cohort study (i.e., Tohoku Study of Child Development, TSCD), focusing both on the potential risks and benefits of fish eating during pregnancy to investigate the effects of toxic chemicals on child development in Japan [
30‐
33]. In this study, we investigated the impacts of PCBs, methylmercury and lead on birth weight in Japanese newborns, with special emphasis on determining whether the effects differed between males and females. At the same time, we reconsidered some confounders that may mislead our result in data analysis.
Results
The mean birth weight of 489 newborns was 3083 (range, 2412–4240) g, and the gestational age was 39.5 ± 1.3 (SD) weeks. The median values of biomarkers in cord blood were 46.0 (5th and 95th percentiles, 18.6 - 113.8) ng/g-lipid for ΣPCB, 10.1 (4.3–22.4) ng/g for THg, and 1.0 (0.6–1.7) μg/dL for lead. The median ΣPCB concentration was 0.122 (0.050–0.332) ng/g-wet and the correlation coefficient (
r) between the wet-base and lipid-base was 0.892 (
p < 0.001). The median THg concentration in maternal hair was 2.0 (0.9–4.4) μg/g and the
r between those in hair and cord blood was 0.841 (
p < 0.001). Since birth weight differed significantly between 252 male and 237 female newborns as shown in Table
1, the following analyses were carried out for the male and female newborns separately.
Table 1
Basal characteristics and exposure levels in 489 mother-baby pairs
Maternal characteristics |
Maternal age (years) | 31.5 ± 4.3 | 31.3 ± 4.4 | 0.497 |
Body weight before pregnancy (kg) | 53.1 ± 7.4 | 52.6 ± 8.4 | 0.511 |
Body mass index before pregnancy (kg/m2) | 21.0 ± 2.4 | 20.9 ± 3.3 | 0.906 |
Body weight gain during pregnancy (kg) | 9.6 ± 3.7 | 9.8 ± 3.6 | 0.584 |
Fish/seafood intake during pregnancy (g/day) | 51.6 (13.1–109.2) | 53.6 (12.2–131.4) | 0.201 |
Smoking habit during pregnancy (smokers, %) | 17 (6.7) | 20 (8.4) | 0.499 |
Drinking habit during pregnancy (drinkers, %) | 82 (32.5) | 73 (30.8) | 0.698 |
Delivery type (spontaneous, %) | 176 (69.8) | 178 (75.1) | 0.225 |
Newborn characteristics |
Gestational age (weeks) | 39.5 ± 1.3 | 39.6 ± 1.2 | 0.516 |
Birth weight (g) | 3126 ± 353 | 3036 ± 314 | 0.003 |
Parity (first-born children, %) | 133 (52.8) | 123 (51.9) | 0.857 |
Biomarkers in cord blood |
Total PCBs (ng/g-lipid)b
| 49.4 (16.3–118.6) | 44.6 (19.6–107.5) | 0.170 |
PCB-153 (ng/g-lipid)b
| 10.6 (3.2–27.9) | 9.7 (3.9–23.4) | 0.243 |
Total mercury (ng/g) | 10.2 (4.5–23.8) | 9.9 (4.0–21.4) | 0.669 |
Lead (μg/dL) | 1.0 (0.6–1.7) | 1.0 (0.5–1.7) | 0.252 |
Table
2 represents correlations between either birth weight or body weight gain during pregnancy and the related indicators. In the 489 mother-newborn pairs, gestational age, BMI before pregnancy, and body weight gain during pregnancy showed significant correlations with birth weight; similarly, maternal age, gestational age, and BMI had significant correlations with body weight gain during pregnancy. Notably, birth weight and body weight gain were negatively correlated with ΣPCB and PCB-153 in the male and female newborns, and THg was inversely related to birth weight only in the male newborns.
Table 2
Correlation coefficients (r) between birth weight and gestational weight gain and the relating indicators
252 Male newborns |
Gestational age | 0.413 | <0.001 | 0.145 | 0.021 |
Maternal age | −0.022 | 0.722 | −0.278 | < 0.001 |
Body mass index before pregnancy | 0.175 | 0.005 | 0.059 | 0.354 |
Body weight gain during pregnancy | 0.200 | 0.001 | - | |
Fish/seafood intake | 0.045 | 0.473 | 0.029 | 0.651 |
Total PCBs in cord blooda,b
| −0.252 | < 0.001 | −0.334 | < 0.001 |
PCB-153 in cord blooda,b
| −0.247 | < 0.001 | −0.325 | < 0.001 |
Total mercury in cord bloodb
| −0.130 | 0.039 | −0.007 | 0.917 |
Lead in cord bloodb
| −0.025 | 0.698 | 0.037 | 0.560 |
237 Female newborns |
Gestational age | 0.311 | < 0.001 | 0.116 | 0.074 |
Maternal age | 0.081 | 0.212 | −0.127 | 0.050 |
Body mass index before pregnancy | 0.176 | 0.006 | −0.245 | < 0.001 |
Body weight gain during pregnancy | 0.251 | 0.004 | - | |
Fish/seafood intake | 0.048 | 0.465 | −0.010 | 0.874 |
Total PCBs in cord blooda,b
| −0.170 | 0.009 | −0.270 | < 0.001 |
PCB-153 in cord blooda,b
| −0.179 | 0.006 | −0.264 | < 0.001 |
Total mercury in cord bloodb
| −0.023 | 0.720 | 0.039 | 0.555 |
Lead in cord bloodb
| −0.030 | 0.642 | 0.126 | 0.053 |
Total newborns |
Gestational age | 0.361 | < 0.001 | 0.133 | 0.003 |
Maternal age | 0.029 | 0.519 | −0.208 | < 0.001 |
Body mass index before pregnancy | 0.171 | < 0.001 | −0.103 | 0.022 |
Body weight gain during pregnancy | 0.190 | < 0.001 | - | |
Fish/seafood intake | 0.043 | 0.348 | 0.009 | 0.843 |
Total PCBs in cord blooda,b
| −0.207 | < 0.001 | −0.307 | < 0.001 |
PCB-153 in cord blooda,b
| −0.209 | < 0.001 | −0.299 | < 0.001 |
Total mercury in cord bloodb
| −0.074 | 0.100 | 0.014 | 0.763 |
Lead in cord bloodb
| −0.022 | 0.635 | 0.078 | 0.086 |
Relations of prenatal biomarkers to birth weight in the mother-newborn pairs are listed in Table
3. In this table, although parity was significantly associated with birth weight in the total newborns, it was not so in the male or female newborns. In adding two interactive variables of (sex) × (ΣPCB) and (sex) × (THg) in independent variables of the total newborns, the two variables were not significantly related to birth weight (
p = 0.983 and
p = 0.373, respectively) but the significance of other variables presented in Table
3 remained unchanged.
Table 3
Relations of prenatal biomarkers and confounders to birth weight: results of multiple regression analysisa
Total PCBs in cord blood | −0.171 | 0.004 | −0.166 | 0.009 | −0.161 | < 0.001 |
Total mercury in cord blood | −0.120 | 0.042 | −0.042 | 0.499 | −0.078 | 0.061 |
Lead in cord blood | 0.023 | 0.692 | −0.039 | 0.513 | −0.011 | 0.784 |
Possible confounders |
Gestational age | 0.397 | < 0.001 | 0.373 | < 0.001 | 0.383 | < 0.001 |
Parity | 0.075 | 0.201 | 0.102 | 0.116 | 0.093 | 0.030 |
Body mass index before pregnancy | 0.143 | 0.012 | 0.227 | < 0.001 | 0.180 | < 0.001 |
Smoking during pregnancy | −0.066 | 0.237 | 0.080 | 0.185 | 0.006 | 0.889 |
Drinking during pregnancy | 0.001 | 0.986 | 0.019 | 0.756 | 0.010 | 0.812 |
Fish/seafood intake | 0.080 | 0.168 | 0.067 | 0.288 | 0.069 | 0.099 |
Child sex | - | - | - | - | 0.159 | < 0.001 |
Contribution rate, R
2
| 0.229 | < 0.001 | 0.204 | < 0.001 | 0.239 | < 0.001 |
As shown in Table
3, lower birth weight was significantly associated with ΣPCB in the male and female newborns, but also with THg only in the male ones. On the other hand, there was no significant association between cord-blood lead and birth weight for either sex. Even when adding body weight gain during pregnancy to independent variables of the multiple regression models, the significance of ΣPCB and THg was almost the same as in Table
3. Concerning the collinearity among exposure markers of 489 newborns, the correlation coefficients were 0.171 (
p < 0.001) for ΣPCB and THg, 0.069 (
p = 0.128) for ΣPCB and fish/seafood intake, and 0.239 (
p < 0.001) for THg and fish/seafood intake; and, the VIFs for ΣPCB, THg, and fish/seafood intake were 1.045, 1.097, and 1.078, respectively.
Discussion
Given the concurrent exposure model of environmental pollutants, the principal findings of our study were that birth weight was associated with ΣPCB concentrations in the male and female newborns, and also with THg concentrations in the male ones. However, the significant relation to lead was not seen, probably not only because of low lead levels but also the extremely narrow range of lead exposure, as compared to the values reported previously [
2‐
6]; whereas, Taylor and coworkers could not find any evidence suggesting a dose–response relationship either in 4190 births (median of maternal blood lead, 3.40 μg/dL; range, 0.20 - 19.14 μg/dL) [
40]. In addition, no evidence of multicollinearity problem among ΣPCB, THg, and fish/seafood intake or of interaction between sex and toxic substances on birth weight was suggested, because all the VIFs were less than 2 in the present study [
39]. Thus, a concurrent exposure model appears to be important for the assessment of the effect on fetal growth/SGA, whereas this model did not include interactive variables between these chemicals. Otherwise, the effect of prenatal exposures to a plural pollutant at low levels may be underestimated or ignored.
In the current study, there was a significant negative relationship between cord-blood ΣPCB and birth weight in the 489 newborns, along with in each group of the male and female ones. This is consistent with results of many reports [
8‐
11,
41‐
43]. Two previous studies observed the negative association only in male newborns [
28,
29], and Yamashita and Hayashi confirmed it only in female newborns [
44]. However, some reports could not find such a significant association [
13‐
17]. At least, all studies using cord blood as a biological specimen demonstrated the significant association; that is, it may be preferable to use the more direct specimen when compared to maternal blood or milk, inasmuch as maternal and umbilical-cord concentrations in red blood cells do not show a significant correlation for several chemical substances [
45]. Taken together, our data support the idea that PCBs can affect fetal growth. The ultimate way to protect fetuses against the harmful effect of PCBs, therefore, is for girls and women not to eat specific fish/marine mammals containing high-level PCBs, like whale blubber, until they have given birth as recommended in dietary advisories of the Faroe Islands [
46].
A significant association of cord-blood THg with low birth weight was observed in the male newborns, but not in the female ones, while the impact of methylmercury did not seem to be stronger than that of PCBs judging from
β values of Table
3. This finding is similar to that reported by Foldspang and Hansen [
18], who examined 376 mothers living in Greenland and demonstrated a negative effect of dietary methylmercury on intrauterine growth. By contrast, Obi and coworkers reported that cord-blood mercury was positively associated with birth weight and length [
47], which may have been due to the fact that their THg levels were extremely low (median, 4.9 μg/L) and their data were unadjusted for any confounders. However, some studies did not detect any significant relationships between the exposure and birth outcomes [
5,
19‐
21]. If these analyses were done for males and females separately, they might have discovered a significant association. Thus, male birth weight appears to have been affected by prenatal methylmercury exposure at relatively low levels (median, 10.2 ng/g in cord blood). In fact, when the methylmercury pollution in Minamata, Japan, was most severe (i.e., in 1955–1959), decreases in male births were observed in the overall city population, in fishermen, and in maternal Minamata disease patients, and an increase in the proportion of male stillborn fetuses was seen [
48]. In other words, male fetuses are suggested to be more susceptible to methylmercury than female counterparts. Sex difference should be kept in mind when assessing reproductive toxicants.
In examining the relationships between environmental pollutants and fetal growth, gestational weight gain is frequently used as a possible confounder [
12,
20,
43]. However, body weight gain during pregnancy was significantly correlated with birth weight (Table
2), and it showed close relations to the ΣPCB and PCB-153, as well as gestational age, maternal age, and BMI before pregnancy, which implies that birth weight and gestational weight gain together are outcome variables originating from intrauterine PCB exposure. Verner and coworkers noted that gestational weight gain is an imprecise measure of the increased maternal lipid compartment during pregnancy [
43]. A significant association of parity with PCB concentrations is also well-known [
35]. In fact, the addition of either gestational weight gain or parity in multiple regression analysis did not change our results when PCB levels existed (Table
3). Accordingly, it is suggested that superficial confounders had better to be excluded from explanatory variables of birth size to avoid misleading or overadjustment.
There may have been some limitations in this study. Although
n-3 PUFAs are essential for normal brain development, we could not determine the levels because of the limited amount of cord blood available. Instead of
n-3 PUFAs, we used fish/seafood intake in the data analysis, whereas it is thought to indicate the exposure levels of both beneficial nutrients and toxic substances such as methylmercury and dioxin-like PCBs [
27,
49]. Other persistent organic pollutants such as dioxin-like PCBs, polychlorinated dibenzo-p-dioxines, and polychlorinated dibenzofurans, were not analyzed in this study but a significant correlation between dioxin-like PCBs and ΣPCB (
r = 0.91) was observed in 49 samples of our participants [
35]. Likewise, smoking habit during pregnancy was employed as a substitute for the number of cigarettes smoked. There is also a paper describing that birth weight may decrease at high intake levels of marine fats [
50], but neither fish/seafood intake nor smoking habit was significantly related to birth weight in the multiple regression analysis as shown in Table
3. As mentioned above, the ΣPCB concentrations in cord blood were determined at two different laboratories, but these values were used following adjustment for the institute. Thus, our data appear not to have been heavily influenced by confounders or measurement bias.
Acknowledgments
We thank all the families for their participation in the cohort study. We would like to acknowledge Drs. Tomoko Sugawara-Oka, Kozue Yaginuma-Sakurai, Takashi Ohba, Satomi Kameo, and Tomoyuki Nakamura, who were members of the TSCD. This research was funded by the Japan Ministry of Health, Labor and Welfare, the Ministry of the Environment, and the Ministry of Education, Culture, Sports, Science and Technology (Grant-in-Aid for Young Scientists [B]).