The strength of the present study is that a population-based birth cohort was followed prospectively with repeated assessment of exposures to marine contaminants for comparison with the allergy and sensitization status up to 7 years of age. Compared with other populations, average exposures to both PCBs and methylmercury were high and ranges of exposures were wide, thereby adding statistical power to the study. Almost 90% of the children participated in the examinations at 5 and/or 7 years of age, and we obtained IgE data at age 7 years from 71%.
As a main finding, serum PCB concentrations at 7 years of age were positively associated with total IgE concentrations. We observed a similar tendency for the prenatal methylmercury exposure, although this correlation could be due to chance. Longer duration of breast-feeding also appeared to predict a higher IgE concentration at 7 of age, but adjustment for the effect of PCB exposure reduced this association so that a chance finding could not be ruled out. For the grass-specific IgE concentration, the duration of breast-feeding again showed a positive correlation, now without a concomitant association with PCB. In addition, we observed an inverse association between grass-specific IgE levels and prenatal methylmercury exposure. Regarding clinical diagnoses, prenatal PCB exposures were inversely associated with a history of atopic dermatitis but showed a weak positive association with asthma. These diverse findings suggest that mechanisms for immunotoxicant effects for total and grass-specific IgE differ from those for asthma and atopic dermatitis.
Because we based the clinical assessment of atopic disease only on examinations at ages 5 and 7 years and on maternal interview, the present study cannot elucidate the possible role of immunotoxicants in the complex pathophysiological origins of these conditions. However, IgE concentrations have become routine clinical parameters in allergological diagnostics (Weinmayr et al. 2010
). We chose total IgE as a marker of general IgE synthesis, and IgE specific to grass (Phleum pretense
) as an important marker of sensitization, because grass pollen is ubiquitous and has previously been demonstrated to be the most common allergen in other North Atlantic environments (Clausen et al. 2008
; Krause et al. 2002b
). For grass pollen, we found 39 positives, corresponding to a sensitization rate of 8.3%, which is slightly lower than found in Western Greenland at 5–18 years of age; this difference is consistent with the lower age of our study population.
Markers of allergic reactions have previously been reported to be associated with a variety of environmental factors (Nagayama et al. 2007
; Poulsen and Hummelshoj 2007
; ten Tusscher et al. 2003
). Sensitization reflected by a specific IgE is not necessarily governed by the same mechanisms as the ones determining total IgE level, as suggested by studies of parental smoking that showed associations in different directions for total IgE and skin test positivity (Strachan and Cook 1998
). Among indications of immune dysfunction associated with increased exposures to PCBs and dioxins, mononuclear cells from cord blood showed decreased in vitro
secretion of tumor necrosis factor-α after mitogenic stimulation; this cytokine is an important proinflammatory stimulant (Bilrha et al. 2003
). Other exposure-related associations include differences in lymphocyte population ratios in peripheral blood from populations exposed to PCBs and related substances (Nagayama et al. 2007
; ten Tusscher et al. 2003
; Van Den Heuvel et al. 2002
). Laboratory animal studies suggest that mercury compounds may induce autoimmune disease and increases in interleukin-4 (IL4) production and IgE levels in certain rodent strains (Nielsen and Hultman 2002
). In human peripheral blood mononuclear cells in vitro
, methylmercury concentrations of 100 μg/L were capable of inducing Th2 cytokine production, whereas γ-interferon production suppression occurred at 400 μg/L; in this model, mercury chloride stimulated increases in IL4 only at 1,000 μg/L (de Vos et al. 2007
In the present study, mutual correlations between PCB and methylmercury concentrations were weak and did not prevent characterization of their differing associations with the immunology parameters. In contrast, individual PCB congeners and ∑PCB concentrations in serum correlated very closely with one another. Close correlation also occurs with other persistent organic pollutants, such as p
′-DDE) (Heilmann et al. 2006
). Although we focused on the ∑PCB concentration as a reliable overall marker of lipophilic contaminant exposure, we were unable to assess possible immunotoxic effects of individual PCB congeners or associated pollutants, which commonly occur in seafood together with PCBs.
Regarding other seafood constituents, maternal n-3 fatty acid intake from fatty fish is thought to affect the development of her child’s immune system (Furuhjelm et al. 2009
). In the Faroese fishing community, positive correlations occur between serum concentrations of n-3 fatty acids and PCBs, although the latter mainly originates from pilot whale blubber (Steuerwald et al. 2000
). Although n-3 fatty acids were not measured in the present study, the absence of any association between maternal fish intake during pregnancy and the immune parameters examined would argue against any important confounding due to maternal n-3 fatty acid intake during pregnancy. In a wider sense, confounding from other risk factors would likely be limited in this Nordic population with relatively uniform living circumstances, also taking into account the increased average level and wide range of exposures to the seafood contaminants (Longnecker et al. 2003
). Thus, the present study considered a substantial number of social and obstetric variables as cofactors, none of which affected our findings. Our results therefore suggest that recommendations on marine food during pregnancy should take into consideration the possible immunotoxic impact of the contaminants.
The observed associations with developmental exposures to suspected immunotoxicants must be evaluated in light of the increased vulnerability of the developing immune system (Dietert 2008
). Because of the semiallogeneic pregnancy state, where graft rejection is suppressed, certain types of effects are more likely to be results of immunotoxicant exposures during the intrauterine developmental phase. During the early postnatal period, both immunosuppression and an increased risk of allergic disease can occur. The last-trimester fetus and the neonate usually exhibit comparatively depressed Th1-dependent functions, and current epidemiologic and experimental evidence on increased total IgE levels after exposure to various forms of stresses suggests that the postnatal acquisition of needed Th1 capacity could be a highly vulnerable target (Dietert 2008
; Poulsen and Hummelshoj 2007
). Accordingly, both dysfunction and misregulation are possible effects of developmental immunotoxicity (Dietert 2008
Although breast-feeding appeared to be positively associated with the total IgE concentration, the adjustment for PCB exposure attenuated this association to a nonsignificant level. The possible impact of lactational immunotoxicant exposure, as reflected by the postnatal serum PCB concentrations in the present study, would suggest that associations between breast-feeding and serum IgE concentrations in children could, at least in part, be due to immunotoxic food contaminants transferred via human milk.
Current evidence is equivocal concerning the effect of breast-feeding on the child’s total serum IgE concentration. A prospective study in the United States found lower IgE concentrations at 8 years of age in breast-fed children, but only if the mother did not have an increased IgE level herself (Wright et al. 1999
). Further, in 258 Pakistani children 6 months to 12 years of age, a total IgE concentration above a reference level occurred in about 80% of bottle-fed children and only in half as many of those that were breast-fed (Satwani et al. 2009
). However, in 215 Polish children 8 months to 18 years of age, the duration of breast-feeding was not associated with the total IgE concentration (Daniluk et al. 2008
Breast-feeding has often been considered a preventive factor in regard to allergy development, although some studies have suggested that breast-feeding may instead cause an increased risk (Bergmann et al. 2002
; Kirsten 2009
; van Odijk et al. 2003
). Also, another Nordic study recently reported that longer breast-feeding was associated with a higher risk of atopic dermatitis but a lower risk of asthma (Giwercman et al. 2010
). The conundrums of statistically significant associations in opposite directions in different populations may be attributable to the effects of one or more independent risk factors that differ between the populations studied. However, data on immunotoxicant exposures are not available from the studies on breast-feeding regarding allergy development or serum IgE concentrations. The present study indicates that an effect of breast-feeding per se is likely to be small and may be negligible. Adjustment for lactational exposures to the immunotoxicants would seem necessary to assess the true magnitude of an independent effect of breast-feeding on allergy risks.
The associations of PCB and methylmercury exposures with indicators of allergy and allergic disease may involve both stimulation and inhibition of immune system functions. Based on the exposure assessments at three or four occasions, prenatal and postnatal exposures seem to have different effects. For methylmercury, we observed exposure-associated effects only in relation to prenatal exposures, and the much lower postnatal exposures did not reveal any clear associations. However, PCB and methylmercury may well target different components of the immune system, and their effects would also depend on the stage of development. Ideally, immune system dysfunction should therefore not be assessed by means of a single or a few parameters nor at one stage of development only.
Our results may not necessarily be at odds with the “hygiene” hypothesis, which has been expressed in different terms regarding allergy and other diseases (Bach 2002
). Rather, our data emphasize the need not to limit the focus only to gene–microbiome interactions but also to include environmental factors, such as immunotoxicants. Because our study provides evidence from a unique population with a well-characterized exposure to environmental chemicals from traditional food, the results provide insight into the potential effects of methylmercury and PCB exposures and their possible interaction with beneficial effects from breast-feeding. Even though the exposure in the Faroes may be less complex than elsewhere, the picture remains multifaceted.
Developmental immunotoxicity may predispose children to common diseases of increasing prevalence, such as childhood asthma and allergic diseases, and is therefore important from a public health perspective. Thus, our findings support the need for screening studies to identify immunotoxicants (Dietert 2008
). In this regard, immunosuppression should not be considered as the only relevant outcome, and effects associated with developmental exposures need to be considered independently from effects in mature organisms. Because of uncertainty regarding interpretation of results from different animal models, human studies remain crucial, and prospective studies must incorporate delayed adverse outcomes of developmental exposures.