Organochlorine pesticides are contaminants that persist in the environment and act as endocrine disruptors in the human body by mimicking, blocking, or interfering with the actions of endogenous hormones (Gore, Chappell et al. 2015). Prior to being banned from use, organochlorines were heavily used internationally to interrupt malaria transmission and for agricultural purposes. Extended use of organochlorines such as p,p’-dichlorodiphenyltrichloroethane (DDT), its metabolite dichlorodiphenyldichloroethylene (DDE), heptachlor epoxide, hexachlorobenzene (HCB), β-hexachlorocyclohexane, oxychlordane, and trans-nonachlor is particularly relevant to human health due to the persistence of these organochlorines in the environment, bioaccumulation in the food chain, and their ability to accumulate in human adipose and other tissues. Organochlorine pesticides have been detected in serum from a majority of persons in the United States and globally and are especially high among adults aged 60 and older (Jaga and Dharmani 2003, Patterson, Wong et al. 2009, Martinez, Trejo-Acevedo et al. 2012, Porta, Pumarega et al. 2012, Sjödin, Jones et al. 2014).

Evidence of sex differences in chronic diseases such as diabetes and kidney disease (Okada, Yanai et al. 2014, Shen, Cai et al. 2017, Madrigal, Ricardo et al. 2019) suggest that perturbations in sex hormone levels may influence disease risk. However, studies of associations between organochlorine pesticide exposures and sex hormone levels in adult males from the general population are limited. Among 178 males living near the United States Great Lakes with measurements of DDE, testosterone, and sex hormone binding globulin (SHBG), no correlation of exposure with hormone levels was observed (Persky, Turyk et al. 2001). However, in a subsample of 56 male participants from the Great Lakes cohort, DDE levels were inversely associated with estrone sulfate (Turyk, Anderson et al. 2006). In a pooled study of males living in Greenland, Poland, Sweden, and Ukraine, DDE was positively associated with SHBG only among the Ukrainians (Giwercman, Rignell-Hydbom et al. 2006), but HCB was positively associated with SHBG among males from Ukraine and Poland (Specht, Bonde et al. 2015). In a sample of 341 males in the United States, no association was observed between HCB nor DDE with SHBG, testosterone, or estradiol (Ferguson, Hauser et al. 2012). There are several studies that have examined the association of organochlorine pesticides with sex hormones among males who have been occupationally exposed or live in highly contaminated areas. Some have reported null findings (Martin, Harlow et al. 2002, Cocco, Loviselli et al. 2004), but others have observed conflicting associations. For example, among 304 males from a rural area of Brazil that was highly contaminated with organochlorine pesticides, heptachlor and DDT concentrations were inversely associated with testosterone levels (Freire, Koifman et al. 2014). However, among 97 males in Thailand, of whom 63.9% had used DDT for farming, neither p,p’-DDE nor p,p’-DDT were associated with testosterone levels but p,p’-DDE levels were inversely associated with plasma estradiol levels after adjustment for age and body mass index (BMI) (Asawasinsopon, Prapamontol et al. 2006).

To the best of our knowledge, the association between organochlorine pesticides and concentrations of sex hormones has not been investigated using the National Health and Nutrition Examination Survey (NHANES). The purpose of this study is to investigate the associations between multiple organochlorine pesticide exposures and sex hormones including testosterone, free testosterone, estradiol, free estradiol, androstanediol glucuronide, and SHBG in a nationally representative sample of adult males in the United States. We explored age, BMI, and prevalent diabetes as effect modifiers of these associations to evaluate if associations might be stronger among certain subgroups. We expected that older participants would have higher levels of organochlorine pesticides and that age would be inversely associated with hormone levels, suggesting that pooled associations might mask age-specific associations. BMI was considered as a potential effect modifier since it is a proxy for fat mass and organochlorine pesticides are stored in fat, which could potentially impact sex hormone metabolism in this tissue compartment. Some organochlorine pesticides have been associated with diabetes prevalence in prior studies, therefore we hypothesized that associations might differ due to different etiologic mechanisms among those with and without diabetes (i.e., that we might only observe associations of organochlorines and hormones among males with diabetes).

In this study, we evaluated the associations between organochlorine pesticides measured in serum and sex hormone levels in adult males. Organochlorine pesticides were detectable in most males in our sample, many of whom had high exposure levels to multiple organochlorine pesticides. We observed associations primarily among HCB and measures of estrogen not only in individual models, but in models adjusted for other pesticide exposures. High concentrations of serum HCB were positively associated with total and free estradiol and inversely associated with the ratio of testosterone to estradiol, independent of the other pesticide co-exposures. In individual models, we observed positive associations among DDT and measures of estrogen, and an inverse association between DDT and SHBG. We observed a suggestive positive association between β-hexachlorocyclohexane levels and estradiol. After adjusting for other co-exposures using principal components analysis methods, the β-hexachlorocyclohexane, DDT, and DDE factor was not strongly associated with any outcome.

Reports that evaluate the association between organochlorine pesticide levels and objectively measured hormones levels in males in the general population are limited. In contrast with our study, HCB was not significantly related to SHBG or estradiol in a sample of males presenting to an infertility clinic (Ferguson, Hauser et al. 2012), but was positively associated with SHBG and free androgen index in males of reproductive age (mean age 29 years) using data from multiple combined cohorts (Specht, Bonde et al. 2015). Consistent with the findings in the current study, most other studies in males have not found HCB to be related to testosterone (Hagmar, Bjork et al. 2001, Goncharov, Rej et al. 2009, Ferguson, Hauser et al. 2012, Freire, Koifman et al. 2014). In non-human studies, effects of HCB on endogenous hormones vary by sex, dose, and species under study. In male rats, there is evidence that at low levels HCB enhances androgen action, but that at high levels it decreases androgenicity (Ralph, Orgebin-Crist et al. 2003).

Interference with normal physiological hormone actions could take place via interaction with the hormone receptor or serum binding proteins, inhibition of enzymes that synthesize hormones, and/or induction of enzymes that metabolize hormones. The mechanism by which HCB could increase estradiol levels is largely unknown. Based on the steroidogenic cascade, the increased serum estradiol levels observed among males with the highest measured HCB concentrations should be coupled with reduced testosterone levels if the substance acting to increase estradiol is up-regulating aromatase, which converts androgens to estrogens. However, the aromatase-driven conversion of androgen to estrogen may utilize a quantitatively small amount of androgen relative to the resulting amount of estrogen (Blakemore and Naftolin 2016). This may explain why we did not observe an association with testosterone despite the observed positive associations with estrogen. The effects of HCB exposure have been compared to those of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), in that HCB binds to the aryl hydrocarbon receptor (AhR) with an affinity approximately 10,000 times lower than that of TCDD (Hahn, Goldstein et al. 1989, Kafafi, Afeefy et al. 1993), in the same range as mono-ortho-PCBs 105, 118, and 156 (Hahn, Goldstein et al. 1989, van Birgelen 1998). Binding of HCB to the AhR could interfere with steroid hormone regulated responses, which could be relevant to the associations we observed. The increased levels of estradiol observed with increasing HCB concentration could also be the result of inhibition of the enzymes that metabolize estradiol through conjugation. This would be consistent with prior studies showing that different types of persistent pollutants may inhibit estradiol sulfation (Kester, Bulduk et al. 2000, Kester, Bulduk et al. 2002, Parker, Squirewell et al. 2018).

Relationships of DDT and its metabolite DDE with endogenous hormones in males have been variable. Consistent with the current study, previous studies have, in general, not found that DDE is significantly related to testosterone (Hagmar, Bjork et al. 2001, Persky, Turyk et al. 2001, Martin, Harlow et al. 2002, Cocco, Loviselli et al. 2004, Asawasinsopon, Prapamontol et al. 2006, Turyk, Anderson et al. 2006, Goncharov, Rej et al. 2009, Langer, Kocan et al. 2010, Haugen, Tefre et al. 2011, Blanco-Munoz, Lacasana et al. 2012, Ferguson, Hauser et al. 2012, Emeville, Giton et al. 2013, Freire, Koifman et al. 2014). One study of males living in a highly contaminated rural area observed an inverse association of testosterone with o,p’ DDT (Freire, Koifman et al. 2014). A cross-sectional study of 50 South African malaria control workers observed positive associations of p,p’-DDT with both estradiol and testosterone, but no association with p,p’-DDE (Dalvie, Myers et al. 2004). In a study of Inuit and three European cohorts, p,p’ DDE was positively associated with SHBG among males from Kharkiv, Ukraine, but not in the other sites and positive associations of DDE with free testosterone were seen in Kharkiv and Greenland but not in the other sites (Giwercman, Rignell-Hydbom et al. 2006). In a South African population with high DDE exposure, positive associations were found with testosterone and free testosterone, as well as with estradiol (Bornman, Delport et al. 2018). In contrast, no associations of DDE with estradiol were seen in populations of Mexican flower growers (Blanco-Munoz, Lacasana et al. 2012) nor in a heavily exposed population in South Africa (Bornman, Delport et al. 2018). In a cohort of Great Lakes anglers, DDE was not associated with SHBG, nor with SHBG bound testosterone (Persky, Turyk et al. 2001, Turyk, Anderson et al. 2006). In a subgroup of males from the Great Lakes cohort, DDE was inversely associated with estrone sulfate (Turyk, Anderson et al. 2006) but the association was not observed in the larger cohort (Persky, Turyk et al. 2001); estradiol was not measured in these studies.

There is limited data on the effects of other persistent organochlorine pesticides on endogenous hormones. Testosterone was inversely associated with β-hexachlorocyclohexane in two studies, but the relationship was not significant among those occupationally exposed (Tomczak, Baumann et al. 1981) and was only of marginal significance among males living in a contaminated rural area with high levels of exposure (Freire, Koifman et al. 2014). Neither serum concentrations of trans-nonachlor nor mirex were found to be related to testosterone in a previous studies of adult males from a highly exposed rural area in Brazil (Freire, Koifman et al. 2014).

In our study, we did not observe strong evidence of effect modification by age, BMI, or diabetes status for any of the associations that we evaluated. Of all the associations we evaluated, we observed one inverse association of heptachlor epoxide and SHBG among males with the lowest BMI (<25 kg/m²), and no associations among males in the 25–29.9 or ≥30 BMI categories. Although the cross-product term for p,p’-DDE and diabetes was significant in the model for SHBG, stratum-specific estimates were not substantively different and trends were null. It is possible that the statistically significant cross-product terms observed were due to chance, and that our ability to detect effect modification was underpowered. To our knowledge, subgroup analyses to identify effect modification for the associations of organochlorine pesticides with sex hormones have not been conducted in other studies with which we can make comparisons.

The associations observed may be relevant to understanding the pathogenesis of selected chronic diseases. Steroid hormones may play a role in the disease pathway connecting persistent pollutants to health outcomes such as diabetes, kidney disease, and cancer. In our study, we found particular organochlorine pesticide concentrations to be positively associated with estradiol and inversely associated with SHBG. Estrogen actions are involved in insulin sensitivity and glucose homeostasis (Mauvais-Jarvis, Clegg et al. 2013). In studies of males estradiol has been related positively and SHBG negatively to diabetes (Kim and Halter 2014, Mather, Kim et al. 2015) and higher levels of SHBG have been associated with reduced risk for low estimated glomerular filtration rate (Kim, Ricardo et al. 2019).

Although many of the organochlorine pesticides studied are now banned from use, recent studies have objectively measured organochlorine pesticide concentrations among adults and observed detectable exposure levels (Jakszyn, Goni et al. 2009, Azandjeme, Delisle et al. 2014, Freire, Koifman et al. 2014, Saoudi, Frery et al. 2014). Our findings may be especially relevant in areas where high concentrations of these pesticides were used, in addition to present-day occupational settings such as those where HCB is produced as a byproduct during the manufacture of solvents and pesticides, pulp and paper production, and metal smelting.

There are several limitations in our study. The NHANES provides cross-sectional data, and only one serum sample was taken to measure both the pesticide and hormone levels. We are unable to determine when the body burden of the measured pesticides was acquired, limiting our ability to tailor intervention strategies to limit or prevent exposure. Only 40.1% of males in our study sample had detectable levels of HCB, suggesting that the observed associations with this pesticide should be interpreted with caution and attempts should be made to replicate these findings in other available datasets. Despite an overall large sample size, we may have had limited power to detect effect modification and precisely estimate stratum specific measures of association. It is possible that exposure to other unmeasured pollutants correlated to the pesticides under study could explain the associations we observed in this study. Residual confounding may also be present due to imperfect methods for adjusting for serum lipids. Although our sample is representative of the U.S. population, subpopulations who are highly exposed, such as those with high occupational exposures or that live in areas of high previous exposure, may be underrepresented. Despite these limitations this study is strengthened by the relatively large, nationally representative sample with a variety of potential confounders and objectively measured organochlorine pesticide (exposure) and hormone (outcomes) levels.

In summary, our findings indicate that exposure to organochlorine pesticides, particularly HCB and DDT, may be positively associated with estradiol levels among adult males in the United States. If substantiated, these findings may provide insights into the mechanism and pathology of chronic diseases known or suspected to be associated with organochlorine pesticide exposure.