Persistent organic pollutants and anti-thyroid peroxidase levels in Akwesasne Mohawk young adults

Introduction

Polychlorinated biphenyls (PCBs) and other persistent organic pollutants (POPs), such as hexachlorobenzene (HCB), and p,p’-dichlorophenyldichloroethylene (p,p’-DDE), a metabolite of p,p’-DDT, have been found to elicit a broad spectrum of biologic, metabolic, and immunologic responses (ATSDR, 2000; Hagmar, 2003; Safe, 2001; Safe, 2000). The potential of these POPs to impair immune responses and trigger autoimmune disease is of growing concern. Given the structural similarity to thyroid hormones, PCBs (and/or their biphenylol metabolites) may mimic thyroid hormone action and even modulate the mechanisms and interfere with the binding of these hormones (Langer et al., 2007; Meeker et al., 2007; Zoeller, 2007; Zoeller, 2001), potentially leading to thyroid functional disorders and autoimmune thyroid diseases (Barsano, 1981; Murai et al., 1987).

Autoantibodies to thyroid peroxidase (TPOAb) can attack the thyroid and damage thyroid function (Larsen et al., 2003). While not a test of thyroid function, serum TPOAb titers are useful as a biomarker of thyroid dysfunction, as well as a diagnostic tool for many idiopathic autoimmune diseases such as chronic thyroiditis (Langer et al., 1998; Langer et al., 2007), Hashimoto’s encephalopathy (Mocellin et al., 2007), Grave’s disease (Antonelli et al., 2006), systemic lupus erythematosus (Kohno et al., 1989), and connective tissue disorders such as Sjogren syndrome (Tunc et al., 2004) and rheumatoid arthritis (Lee et al., 2007). The likelihood of damage appears to be greater in females, and increases with age (Antonelli et al., 2006; Lee et al., 2007).

Autoimmune dysfunction is thought to disrupt normal hypothalamic-pituitary-gonadal function and thereby affect reproductive outcomes, such as an increased risk of pregnancy loss (Djursing et al., 1982; Gleicher et al., 1993; Kjaer et al., 1992), and has been linked to the impairment of female fecundity even before the diseases become clinically overt (Nelson et al., 1993). Elevated TPOAb levels have been linked to autoimmune hypo- and hyperthyroidism in pregnancy, immune mediated miscarriages, and an increased risk for postpartum thyroiditis (Adcock et al., 1994; Matarese et al., 2003).

Here we examine the relationship of different PCB groupings, according to chlorination level and substitution pattern, along with p,p’-DDE and HCB, to anti-thyroid peroxidase antibody among young adults of the Akwesasne Mohawk Nation.

Methods

Results

The median age of participants was 17.6 years of age; 48% were females (61 males, 57 females). Forty seven participants were breast fed. The mean duration of breast feeding was 2.5 months (sd = 4.4 months). Thyroid hormone levels are described in Table 2. Nineteen participants (16%) had fT3 levels below the laboratory reference range; none had elevated levels. Of these 19, one participant had both fT4 and T4 levels below the laboratory reference range; one participant had elevated T4 and a TSH level below the reference range, and one had elevated TSH and TPOAb levels, and a T4 level below the reference range. Overall, eighteen participants (15.4%) had TPOAb levels above the normal laboratory reference range (23% of females, 9% of males). Levels of thyroid hormones did not differ by breast feeding history. TSH, T3, and fT4 levels did not differ by sex although Akwesasne females had significantly higher levels of T4 (p<0.0001), while males had higher levels of fT3 (p<0.0001).

Toxicant levels did not differ by gender. Breast fed participants (n=47) had higher levels of HCB and p,p’-DDE than non-breastfed subjects. All PCB groups (PCB50%, PerPCB9, ΣDioPCBs, ΣEnzPCBs, ΣWolff1-3) but one (ΣNONPER5) were significantly higher among breastfed youth. Mirex was unrelated to breastfeeding history (Table 3). Among participants who were breast fed, those with an elevated TPOAb level had significantly higher levels of all PCB groupings, with the exception of levels of ΣNONPER5 which did not differ significantly; levels of p,p’-DDE were also significantly elevated (Figure 1). HCB and mirex (not shown in figure) were not higher among those with elevated TPOAb.

We calculated Pearson correlation coefficients to determine the relationship of toxicants to TPOAb among youth (Table 4). When breast fed and non-breast fed participants were combined, no significant relationships were present (the lowest p value equaled 0.44). However, after stratifying by breast feeding status, participants who were breast fed showed significant, positive relationships between TPOAb levels and these PCB groupings: ΣPCB50%, ΣPERPCB9, ΣMonoPCBs, ΣDioPCBs, ΣTriPCBs, ΣWolff2, ΣWolff3, and ΣEnzPCBs; but not with ΣNONPER5, p,p’-DDE, HCB, or mirex. Multivariate regression analysis was performed (Table 5), examining the effect of each toxicant while including age, sex, number of years of education completed, and current cigarette use. None of these covariates were significant in the model and the results were not different from those obtained with the bivariate analysis except for p,p’-DDE which was found to be significant (p=0.05). Among the non-breast fed group, no significant relationship was found with any of the toxicants in bivariate or multivariate analyses.

Discussion

Our results show that the groups of more persistent PCBs are unambiguously associated with an elevated TPOAb level among late teens and young adults who had been breast fed as infants. Associations with p,p’- DDE are also evident, but these associations are not as consistent. HCB and mirex appear unrelated to elevated TPOAb levels. Consistent with these associations we also observed that within the normal range of TPOAb levels, groups of mono-ortho PCBs were positively associated with TPOAb levels. Furthermore, the relationships appeared in both males and females, further supporting the consistency of association between PCBs and TPOAb level. The finding that these relationships were only evident among breast fed youth suggests that perinatal or postnatal exposure to PCBs is influential. Because breast feeding introduces substantial amount of PCBs to the infant (Hooper, et al, 2007; Nagayama et al., 2007; Patandin et al. 1999), one possibility is that the effect is due to the greater exposure itself. Although the PCB levels of the breast fed and non-breast feed youth differ by at most 17% at the age of examination, the difference was likely much greater during infancy depending on the extent of breast feeding and the maternal body burden (Patandin et al. 1999). Another possibility is that the effect is a matter of timing. When we restricted the analysis of the non-breast fed subsample to those with PCB levels comparable to the breast fed youth, there were no relationships between levels of PCBs and TPOAb. This suggests that the effect on TPOAb may be more one of when exposure occurs rather than how much exposure occurs. It is also possible that breast feeding is associated with other factors that are contributory, but the most likely of these (such as age and gender) were included in the multiple regression analysis and had no effect on the relationship. Unknown, and therefore, unmeasured factors are always a possibility.

The use of multiple groups of congeners was designed to test effects of functionally or structurally defined groups discerned by other investigators even though there is substantial overlap among the congeners contained in several of the groups. It is not possible to discern a clear difference among the PCB groups in terms of the strength of their relationships to elevated TPOAb among breast fed young adults with the exception of the non-persistent PCB congener group and possibly the ΣWolff1 group. Both groups are largely composed of non-persistent congeners (Wolff et al., 1997). Not including those groups, the average difference in PCB level between elevated and not elevated TPOAb groups is 0.24 ppb while the non-persistent group differs by only 0.04 ppb, and ΣWolff1 group by 0.05 ppb. Likewise, the r values and β weights describing the relationships of specific PCB groups to TPOAb levels are of similar strength, except again for the two groups of relatively non-persistent congeners. This suggests that the only characteristic of the PCB groups that is relevant to the effect on TPOAb in this sample is relative persistence. This may suggest that current exposure, which is best reflected by the levels of non-persistent congeners, has little effect on TPOAb levels among young adults, whereas earlier exposure, reflected by serum levels of more persistent congeners, does.

We did consider using a TEQ approach but due to insufficient specimen availability from these participants, we could not obtain data for the most potent dioxin-like congeners (126 and 169). Although we could base a measurement on the levels of the dioxin-like CBs 105, 118, and 156, without data for the most potent CBs, the results of such an analysis would be unreliable. We also considered data analyses based on lipid-adjusted PCB values. However, lack of standardization of laboratory methods and the need for considerable technical expertise in conducting total lipid determination can result in bias and variability in these comparisons (Schisterman et al., 2005; Bernert et al., 2007). Consequently, because we were not performing direct comparisons with other investigators’ data, and since all of our subjects were fasted prior to blood sampling, we decided not to lipid-adjust our PCB data for this study.

Although the associations reported here support a causal relationship, cause itself is not proven; additional support could come from knowledge of a biological mechanism. However, the mechanism of this effect is not clear. Laboratory studies have not investigated TPOAb levels in relation to PCBs, and this would be an important experimental approach. There are indications that PCBs are associated with frank autoimmune diseases, such as diabetes (Codru et al., 2007; Fierens et al., 2003; Lee et al., 2006), rheumatoid arthritis (Lee et al., 2007), and autoimmune thyroiditis (Langer et al., 2007; Langer et al., 2003). Our results for PCBs agree with those of Langer et al. (2007), who found a relationship between TPOAb levels and levels of several organochlorines (PCBs, HCB and p,p’- DDE) among adults with “long term high exposure.” One feature of our study is that levels of PCBs are considerably lower than those found in the study by Langer et al. (2007), and in many other studies of adolescents and young adults (Chevrier et al., 2007; Osius et al., 1999). This suggests that PCB exposures at levels commonly found in the US population and elsewhere could promote autoimmune disease if exposure occurred during a critical window as is suggested here. Considering these findings, we conclude that exposure to background levels of PCBs may in fact increase the risk of acquiring an autoimmune disease. There may be multiple mechanisms involved in the initiation and progression of autoimmune diseases by toxicant exposures as modified by individual differences in metabolism and lifestyle. Further studies are necessary to elucidate the site and mechanism of action of these POPs and to establish thresholds for these effects, especially among populations with background levels of toxicant exposure.