Vreugdenhil, HJI, FME Slijper, PGH Mulder, and N Weisglas-Kuperus 2002. Effects of Perinatal Exposure to PCBs and Dioxins on Play Behavior in Dutch Children at School Age. Environmental Health Perspectives 110:A593-A598.

In this groundbreaking study, Vreugdenhil et al. report that perinatal exposure to PCBs is associated with changes in play behavior of children. Boys with higher PCB exposures are less likely to engage in masculine patterns of play, and girls with higher exposure are more likely to engage in masculine play. They also find an association with dioxin: more feminized behaviors were found in both boys and girls exposed prenatally to higher levels of dioxin. While this is the first study ever reported of the effects of contaminant-induced endocrine disruption on gender differences in child behavior, its results are consistent with animal studies.

It should be noted that the PCB levels studied here are not terribly high, nor a result of occupational exposures. They are background levels resulting (most likely) from normal consumption of foods from Dutch markets. The study focused on differences among the mothers in their PCB body burdens: while some had less, some had more, none had none nor did any have levels out of the ordinary.

What did they do? Vreugdenhil et al. study continues the work by this team of Dutch scientists on a cohort of children whose PCB and dioxin exposures have been carefully characterized. Previously reported findings include adverse impacts on immune system function and neurological development.

Their research focused on 207 Dutch infant-mother pairs recruited to the investigation between 1990 and 1992. Chemical exposure was assessed at three stages: from mothers' blood drawn during the final month of pregnancy, from umbilical cord blood obtained at birth, and from breast milk within two weeks of birth. In the blood samples, they measured levels of four PCB congeners (118, 138, 153 and 180). Measurements of breast milk were more extensive: 26 PCBs and 17 dioxins. They combined the measurements of the individual congeners using the standard TEQ approach (toxic equivalents).

Only half of the infants were breast fed; the others received infant formula. For the breast-fed infants, total post-natal TEQ was estimated based on the concentrations measured in breast milk and the duration of breast feeding. In combination with the blood samples, these data also allowed the investigators to assess the impacts of exposure after birth.

To characterize play behavior, Vreugdenhil et al. used the Dutch version of the Pre-School Activities Inventory (PSAI), a standard method for assessing gender aspects of play behavior in young children.

According to the authors:

As in their earlier studies, the Dutch scientists used statistical procedures (multiple regressions) to control for the potential influence of confounding variables that might affect neurological development, including education level of parents, verbal IQ of the principal parent, smoking, alcohol use, age of the mother, etc.

The average age of the children at the time of behavioral assessment was seven and a half years.

What did they find? On average, boys measured using the PSAI method scored higher on the masculine scale than girls and lower on the feminine scale.

These gender differences are all significantly different (p < 0.001) and are precisely what is expected: on average, boys engage in more masculine play, girls in more feminine play.

In the multiple regression analysis of contaminant effects, Vreugdenhil et al. found statistically significant relationships between their measures of masculinity and femininity of play behavior and PCB and dioxin exposures.

With increasing prenatal PCB exposure, boys exhibited less masculine play behavior. Their composite scale score rose and their masculine scale score fell. The association was stronger with cord blood PCB levels (p < 0.011 and p < 0.001, respectively, for composite and masculine scales) than with maternal blood during the final month of pregnancy (p < 0.03 for both scales). Girls scores fell on the composite scale and rose on the masculine scale, but the trends were not significant statistically. Feminine scale scores did not change for either boys or girls in relation to prenatal PCB levels.

In contrast, prenatal dioxin levels were associated with higher feminine scale scores in both boys and girls (p = 0.048), "indicating more feminized play behavior in both sexes."

Higher PCB levels in breast milk were associated with more masculinized behavior in girls (p=0.028). The trends for boys suggested association with more feminized behavior but was not significant.

However, controlling for prenatal exposure, there was no link between total post-natal exposure via breast milk and trends in play behavior.

What does it mean? Vreugdenhil et al. interpret their results to indicate that prenatal exposure to PCBs and dioxin interfere with the hormonal signals controlling development of sexual differences in typical play behavior:

They acknowledge that their methods do not allow a complete test of this hypothesis, especially as they have no information on the levels of natural steroid hormones experienced by the children prenatally, nor is there full understanding of the (likely multiple) ways that PCBs and dioxin disrupt steroid hormone action during development.

The results nonetheless are consistent with previous indications that prenatal hormone levels can affect sexual differentiation of childhood play.

Some human data, moreover, already implicate contaminants in disrupting sex-specific behavioral development. Research with babies born to mothers exposed to PCBs and PCDFs in the Yucheng rice oil contamination incident found a diminution of spatial orientation abilities in boys, with no effect in girls. Because boys typically perform better on tests of spatial orientation than girls, these results were interpreted as demasculinizing or feminizing effects caused by disturbances in steroid hormones by prenatal exposure to PCBs/PCDFs.

Vreugdenhil et al.'s results are important for two reasons.

• First, they clearly implicate hormonally-active contaminants in disruption of the development of gender-specific human behavior, even if they fall short of establishing causality.
• Second, the affected behaviors, sex-specific childhood play, has been linked speculatively to later patterns in adult sexual choice. While highly controversial, at least some research (e.g., Green 1985) suggests that "childhood gender nonconformity" (e.g., play behavior typical for opposite sex) is a predictor of same-sex sexual choice in adulthood. Whether or not the observations reported in by Vreugdenhil et al. are relevant to this theory remains to be determined.