Discussion
Accidental and occupational exposure to PCBs has already been associated with increased susceptibility to infections in infants. Rogan et al. (1988) observed that mothers who were exposed to PCBs through the consumption of contaminated rice oil (Yu-Cheng) reported a higher rate of bronchitis in their children than did control mothers. After examination by two otolaryngologists, the same children were also shown to have a higher prevalence of middle ear diseases than matched controls (Chao et al. 1997). In Japan, Hara (1985) noted that infants born to women who had handled PCBs in a capacitor factory had a higher incidence of colds and GI complaints.
However, evidence of an effect of environmental OC exposure on susceptibility to infection in children is scarce and inconsistent. To our knowledge, the first study addressing this question was conducted in the Great Lakes area (Smith 1984); the author observed that fish consumption during pregnancy (a proxy of PCB exposure) was positively associated with colds, earaches, and flu symptoms in infants. Rogan et al. (1987) followed 900 families in North Carolina (USA) between 1978 and 1982. They reviewed children’s medical charts and did not find any evidence of harmful effects of PCBs or DDE during the first year of life. In the Netherlands, Weisglas-Kuperus et al. (1995) observed no association between PCBs and the number of episodes of rhinitis, bronchitis, tonsillitis, and otitis during the first 18 months of life. However, in the same group of children at 42 months of age, current PCB burden was associated with a higher prevalence of recurrent middle ear infections and chicken pox (Weisglas-Kuperus et al. 2000). Karmaus et al. (2001) also observed a higher risk of otitis media, but the association was only present with the combined exposure to DDE and PCBs. Finally, our laboratory previously reported that exposed Inuit infants had a higher risk of acute otitis media during the first year of life (third tertile of exposure compared to the first) (Dewailly et al. 2000). The association was significant with exposure to DDE and HCB but remained above the unity for PCBs, dieldrin, and mirex.
In this study, we showed that prenatal exposure to some environmental OC contaminants was possibly associated with a higher incidence rate of infections during the first 6 months of life. Although the associations were not always statistically significant because of limited statistical power, infants in the highest quartiles of PCB and DDE exposure had systematically more episodes of infections than their counterparts in the first quartile of exposure. This was mostly observed during the first 6 months of life, as the effect size was lower when infections during the first 12 months of life were considered. Postnatal exposure to OCs was not associated with infection incidence.
In the literature, middle ear infections are the most consistently reported infections associated with prenatal exposure to OCs. In our study, the strongest dose–response relationship was seen with ear infections. However, it is likely that insults of OCs on the developing immune system would result in the increase of incidence of many different types of acute infections and not only ear infections. Consistent with that assumption, our results showed a higher incidence rate for the four most frequent infections in infants in the higher exposure groups, and the rate ratios were similar to that observed for otitis. Furthermore, when these four types of infections were combined, the association was more stable and the magnitude of the dose–response relationship was increased, compared with that of the four types of infection taken separately.
We also observed that the effect of pre-natal exposure was mostly present during the first few months of life and that this effect seemed to vanish after 6 months of life. Furthermore, we found no effect of postnatal exposure to OCs with infections. It has already been suggested that the immune system is vulnerable to immunotoxic compounds during its development and that high maternal burden during pregnancy and lactation could lead to permanent defects on the infant’s immune system (Badesha et al. 1995; Barnett et al. 1987). Our results support the hypothesis of a stronger effect during early infancy, but we were unable to clearly identify any harmful effect persisting after the age of 6 months. After a few months of life, cumulative environmental influences on the immune system may begin to play a larger role, thus increasing the variability of responses to infections. Furthermore, contributions of the OC exposure via breast milk, entangled with the beneficial effect of breast-feeding on infections, might have masked the effect. This could explain in part the discrepancies in results of other studies on OCs and infections because the age of children during disease and exposure assessment varied considerably between studies. Further studies are needed to clarify the time period during which environmental exposure to OCs has a detrimental effect on children health.
In this population, plasma concentrations of many environmentally persistent OCs are strongly correlated (Muckle et al. 2001). Muckle et al. (2001) also showed that concentrations in cord plasma, maternal plasma, and breast milk samples are also strongly correlated. With such exposure, it is therefore not possible to attribute the effect observed to one specific OC compound, nor are we able to unravel the specific contribution of PCB-153 exposure from DDE exposure. Furthermore, our data did not allow us to determine whether the association between DDE and infections was due to an immune modulation property of DDE, to co-linearity with PCB-153, or both.
We used a review of the medical charts to evaluate disease frequency. There is only one health center in each of the three Inuit communities included in this study, and participants almost always go to that heath center when they seek medical attention; copies of consultations performed elsewhere are routinely requested to complete medical charts. We are therefore confident that we have reviewed a majority of the medical consultations sought by the participants. Nevertheless, we did not attempt to verify every diagnosis, nor did we try to inquire about infections for which medical attention was not sought by the parents. Furthermore, we did not find a suitable proxy for the propensity to go to the clinic when symptoms were present (health services are free of charge in Canada). Our results are therefore likely to be an underestimation of the true incidence. This underestimation is expected to be present for benign infection, but is unlikely to be significant for LRTIs. This underestimation may be associated with traditional lifestyle, and thus with OC exposure, but the direction of the bias is unknown. However, if such a bias was present, we could assume that it would have persisted beyond 6 months of age. RRs for the 12-month follow-up are close to unity; therefore, the bias seems to have little effect on our results.
Because of the relatively small number of subjects involved (n = 199), our results must be regarded with caution. Many factors can greatly influence the rate of acute infections. We have assessed several potential confounding factors, but unknown factors might still be present. Specifically, we cannot rule out the possibility that the infants in the lowest exposure group (first quartile) had better general health due to an unknown cause or simply due to chance. This would have resulted in RRs above the unity for the three highest quartiles of exposure without any dose–response association, which is similar to what we observed. This should be kept in mind in interpreting our results.
The high rate of infectious episodes in young Inuit children has been observed in northern Canada, the United States (Alaska), and Greenland (Banerji et al. 2001; Holman et al. 2001; Koch et al. 2002; Proulx 1988; Wainwright 1996). Many cultural, environmental, and economical factors contribute to this situation. Our study population is no exception, with a mean of almost nine infection-related medical consultations per infant during the first 12 months of life. In the context of such a high rate of infections, rate ratios of around 1.25, like the ones observed in this study, could have a tremendous impact on the public health of this population. This is the second study identifying a possible association between acute infections and prenatal exposure to OCs in Nunavik. However, the relatively small number of subjects raises the possibility of an association that could be due to chance. To further clarify the potential contribution of persisting contaminants in the high infection rate of these children, we are currently conducting another study in which a third cohort of Inuit children from the same population is being followed during the first 5 years of life. Other studies are also needed to identify which immune mechanisms are involved and to better understand the role of maternal passive immunity in these infants. In the meantime, awareness and precautions regarding the selection of marine food items before and during pregnancies are warranted.