Perfluorinated compounds are related to breast cancer risk in greenlandic inuit: A case control study

Breast cancer (BC) is the most common cancer for women in the western world and the incidence has been increasing since 1940. The highest incidence rates are observed in North America, and the lowest risk is found in Asia and Africa [1]. Breast cancer is also the most common cancer in females in Europe with the highest incidence in The Netherlands and Denmark and lowest in the eastern part of Europe [2]. From very few cases in the 1970's an extraordinary increase in BC has been observed in the Inuit population of Greenland and Canada today [3, 4]. Known established breast cancer risk factors include genetic inheritance e.g. mutations in the BRCA-1 and BRCA-2 genes [5], lifelong exposure to estrogens (early menarche and late menopause increases the risk), obesity after menopause, alcohol, smoking and high intake of fat [2, 6, 7]. Some factors seem to reduce the risk such as low age at first birth, large number of full term pregnancies and long duration of breastfeeding [6, 7]. Thus, BC risk is influenced by genetics and reproductive history, but the known risk factors only explain less than a third of all cases and more than 70% of women diagnosed with BC have no inherited or sporadic cancer. The risk of BC is thought to be modified by lifestyle and environmental exposures. Results from many studies have confirmed that BC is not a single disease with variable morphologic features and biomarkers, but rather a group of molecularly distinct neoplastic disorders [8].

The susceptibility to BC upon environmental exposures might not only be as child, in puberty or young but can be pre-determined in utero through alterations of the hormonal environment caused by either maternal diet and/or exposure to environmental chemicals with endocrine activities that can modify the epigenome. These epigenetic modifications, inherited by somatic daughter cells, lead to changes in mammary gland development and increase the vulnerability for malignant transformation [9]. An array of legacy persistent organic pollutants (POPs) including polychlorinated dibenzodioxins and polychlorinated dibenzofurans (PCDD/PCDF), polychlorinated biphenyls (PCBs), and organochlorine pesticides (OCPs) are potential endocrine disrupters and can play an important role in the risk of BC. These persistent and lipophilic compounds do biomagnify through the food web and bioaccumulate age dependently in human and animals. PCBs have been associated with effects being relevant for development of BC such as estrogenic, tumour promoting, and immunosuppressive activities [10]. Evidence regarding organochlorine exposure and BC risk is controversial. A case-control study of Danish women showed a significant doses relation between the risk of BC and the serum level of the pesticide dieldrin, and a non-significant relation of BC risk at the highest level of dieldrin and PCBs among BC women with a tumour with mutant p53 [11, 12]. In contrast, Raaschou et al. 2005 found no association between persistent organochlorines (14 pesticides and 18 PCBs) levels in adipose tissues and breast cancer in postmenopausal Danish women [13]. Thus, no strong evidence for the role of organochlorines, including PCBs, in BC development was found. However, two studies found a 3-fold risk in postmenopausal breast cancer with a A2455G mutation in the P450 polymorphic CYP1A1 gene and high PCB levels compared with wild-type alleles and low PCBs [14]. These associations between PCB exposure and BC risk among genetically susceptible and racially subgroups warrant further studies.

Perfluorinated compounds (PFCs) are a large group of chemicals used since the 1950's in different industrial and commercial applications (e.g. Teflon, carpets, furniture, food stuff packing etc.). For a long time, these fluorinated chemicals were considered metabolically inert and nontoxic [15] and the carbon-fluorine bond renders theses chemicals very resistant to biodegradation and therefore persistent in the environment [15]. Available evidence suggest the transformation or biodegradation of precursor perfluorinated chemicals occurs by both abiotic and biotic degradation pathways where perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) are typical final degradation products [16, 17]. In 2001, it was discovered that the PFCs were accumulating in the environment and human tissues with a global distribution [16–18]. Unlike the legacy POPs (e.g. PCBs, OCPs) - which accumulate in lipid rich tissues - PFCs bind to blood proteins and accumulate mainly in liver, kidney and bile secrets [19]. Humans are exposed to PFCs through occupational settings, environmental exposures and/or through contact with consumer goods (diet, air, water, food and household dust) where PFCs have been found.

The perfluoroalkyl acids (PFAAs) include the perfluorocarboxylated acids (PFCAs) and perfluorosulfonated acids (PFSAs), and the PFAAs include the two most studied PFCs: PFOS and PFOA. These two compounds are the most studied because existing laboratory procedures in the past did not allow analyses of other PFCs that in general exist in lower concentrations. PFOA and PFOS are persistent in the environment and found in human blood, breast milk and liver with half lives of 4 to 10 years [20]. The PFCs are found globally and governmental regulations in USA and Europe on use and production of specific compounds such as PFOS and PFOA have been made. Recently, PFOS has been added to Annex B of the Stockholm Convention on POPs [21].

Biomonitoring studies have been carried out in almost all part of the world in order to assess PFAAs levels and temporal trends in the general population [22, 23]. European studies observed serum and plasma concentrations range from 1 to 116 ng/ml for PFOS and from 0.5 to 40 ng/ml for PFOA. The average plasma levels of PFOS and PFOA in Danish pregnant women was 35.3 and 5.6 ng/ml, respectively, similar to most levels reported for western country populations during the same decade [24]. For middle aged women in Norway slightly lower levels were reported (medians: PFOS; 20 ng/ml; PFOA; 4.4 ng/ml). Mean and median concentrations from North American populations appear to be slightly higher than European, Asian and Australian populations [22]. Recently we performed a survey of serum PFCs in Greenlandic Inuit [25] showing similar PFOS levels in Inuit women as found for Danish and European women but lower PFOA levels.

The maternal level of PFOA or PFOS was found associated with various reproductive and child health outcomes [24, 26–29], and seems to impact maternal fecundity [27].

The biological effects of PFCs have been studied in more detail mainly in rodents; little data are available for other species and humans [19, 30]. Studies in animals have documented an array of toxicological outcomes including liver hypertrophy and tumours [31], thyroid hormone alterations, decreased serum cholesterol and glucose, developmental toxicity, immunotoxicity, and carcinogenic potency [32, 33]. Animal and in vitro studies have also suggested that PFAAs may have potential geno- and neurotoxic effects [34–36]. U.S. EPA has proposed PFOA to be deemed as a rodent carcinogen with relevance to humans [37].

A 2-year study in rats [38] reported a statistically significant increase in mammary fibroadenomas and Leydig cell adenomas suggesting impact of PFOA on reproductive tissues. Because of these data the U.S. EPA Science Advisory Board recommended to reconsider the possible impact of PFOA on mammary tissues. In 2007 White and coworkers reported [39] that gestational exposure to PFOA in mice compared to non exposed controls was associated with altered mammary gland development in dams and female offspring. A significant reduction in mammary differentiation among exposed dams was evident, and also affected the epithelial involution and altered milk protein gene expression [39, 40].

Recently, estrogen-like properties of PFCs were reported in human MCF-7 breast cancer cells suggesting endocrine potentials [41].

The objective of this BC case-control epidemiologic study in Greenlandic Inuit women was to evaluate the serum level of legacy POPs and PFCs, blood metals, and the combined xenobiotic serum POP related effect on the functions of the estrogen- (ER) and androgen receptors (AR) and the aryl hydrocarbon receptor (AhR) functions.

This study aimed at evaluating the association between serum levels of legacy POPs and PFCs in Greenlandic Inuit breast cancer cases and their controls. The present data shows for the very first time a relation between the serum level of PFCs and the risk of breast cancer in Greenlandic Inuit. Moreover, the breast cancer cases also had a significant higher serum PCB level at the highest quartile. In addition, the cases elicited a higher frequency of subjects samples inducing POP related hormone-like agonistic xeno-ER (XER) and xeno-AR (XAR) transactivity being significantly different for XAR. In contrast, the serum POP related dioxin-like AhR-TEQ was lowest in BC cases, although the significance disappeared for the adjusted data.

Until now very few reports are published concerning the incidence of breast cancer in the Arctic population and to our knowledge none have ever evaluated the factors affecting the risk of breast cancer in Greenlandic Inuit. In the western world BC is the most common cancer for women [1, 2], where established risk factors can explain less than 30% of the cases. The incidence of BC has traditionally been low among the Inuit, but since the 1970's a considerable increase has been observed [3, 4] although still at a level being approximately 60% of the incidence in e.g. Denmark [58]. In the Arctic, before 1966 BC was reported absent from the western and central Canadian Arctic and from 1967 to 1980 only 2 BCs out of 107 cancers were found [59]. Breast cancer was studied over a 20-year period in Inuit populations in the circumpolar region. A total of 193 BC's were observed in women with an incidence increase from 28.2 per 100.000 in 1969-1973 to 34.3 per 100.000 in 1984-1988 [60]. In Greenland, the age adjusted incidence in women increased from 35 to 46.4/100000 in the period 1973-87 and 1988-97, respectively [58]. Among Greenlandic women, an increase is particular observed among older women e.g. at the age from 40 to 70 the increase was (~65/100,000) in 1973 - 87 compared to (~170/100,000) the period 1988-97. [58, 4]. There is a pattern shift typically seen in low-risk countries with stagnation or falling rates after menopause to increased rates after menopause as observed in the Western countries. This pattern shift does not support a general improved diagnosis that would be expected to be reflected across all age groups. Ethnic and international differences in BC risk can mostly be explained by differences in environmental exposures and lifestyle, particularly reproductive and hormonal factors [6–8]. The increasing prevalence of obesity and type-2 diabetes [61], and change in breastfeeding from prolonged feeding continuing through childbearing years to a pattern of breastfeeding now more similar to the western world [4] could be consisting with the increase in Arctic BC rates. However, low parity, late age at first birth and obesity is also risk factors for hormone-related cancers such as uterine and ovarian cancer that have not elicited a parallel increase in incidence in the Arctic and thus other factors may be important for the increase in BC incidence [4].

In this study the median age and BMI were similar for BC cases and the controls and no difference in breastfeeding was found. The only established reproductive BC risk factor found significantly different from the controls was, that BC cases had a lower number of full term pregnancies, although this information was only obtained for approximately 50% and 77% of the cases and control participants, respectively. In contrast to the expected menopausal related BC risk a higher frequency of cases were premenopausal (55% vs. 18%), and for the controls a higher percent of the subjects were postmenopausal. We found that postmenopausal women had higher burden of PFCs and legacy POPs than premenopausal women both in cases and controls, supporting the known age-dependent bioaccumulations of these compounds [25, 43]. It is expected that premenopausal women are at general lower risk for BC than postmenopausal women. However, we observed a higher frequency of premenopausal women in the case group during the sampling period 2000-2003. We suggest that the combined endogenous and xenobiotic related hormone disruption might be higher for cases compared to controls as hypothesized in the following: Premenopausal cases, although relatively shorter lifetime exposure to endogenous estrogens but higher actual estrogen levels, the higher legacy POPs and PFC serum levels contributes to a higher POP related xenobiotic estrogenic disruptions. For the postmenopausal cases, although lower actual level of endogenous estrogen, the BC risk is influenced by a longer lifetime estrogenic exposure and the higher legacy POPs and PFC serum level contributes to a higher POP related xenobiotic estrogenic disruptions. This is supported by our data analysis stratifying the menopause status showing that cases had higher levels of PFCs and legacy POPs than controls both for the premenopausal women as well as postmenopausal women. In addition, postmenopausal women might be at higher risk because of significant higher xenoandrogenic bioactivity in accordance to the hypothesis of Adams, J.B. [62].

Result from the questionnaire showed similar proportion of reported current smoking for cases and controls. However, since questionnaire is subjective, the measurement of serum cotinine is more precise to reflect the current smoking status. The lower level of serum cotinine for cases indicates that the cases smoked less but it can be a result of their disease although the non significant lower level of serum Cd in cases supports different smoking habits for cases.

The Arctic Inuit population have one of the highest burden of POPs globally, particularly in some districts in Greenland, and the hormone disrupting potential of the actual mixture of serum POPs have been reported [63]. The recent increase in BC incidence might be explained by the high burden of legacy POPs and increased exposure to new emerging POPs such as PFCs together with the recent transition in the Inuit diet from the traditional marine food to more western food and lifestyle factors such as smoking and alcohol intake.

The ubiquitous presence of PFAAs globally including the Arctic regions has been documented [64, 25], and unlike legacy POPs the level of PFAAs in marine mammals from Greenland still show an increasing trend in the past years [65, 66].

In the present study we measured 3 PFSA isomers and 7 PFCA isomers and the most abundant isomers were PFOS and PFOA, and to a less extend PFHxS. The statistic data for PFHxS was similar to that of PFOS (data not shown). Since PFOS and PFOA were the most abundant we use these two compounds as the representative isomer of PFSAs and PFCAs. For BC cases we observed a significant higher serum level of PFCs and for PCBs in the highest quartile, and the adjusted odds ratios (ORs) indicated a significant risk for BC in relation to the level of serum PFOS and sumPFSA and total sum of legacy POPs plus PFCs.

Our data shows to our knowledge for the very first time an association between the serum level of PFCs and the risk of breast cancer. The higher levels of PCBs in BC cases in the present study are supported by previous reported data [10–12, 14].

In general, there have been reported non consistent association between serum fluorochemical levels and adverse health effects in human but elevated bladder cancer mortality among male workers exposed to PFOS for a minimum of one year [33]. In rats [38] and mice [39] studies upon gestational exposures to PFOA significant increase in mammary fibroadenomas and Leydig cell adenomas and significant reduction in mammary differentiation and gland development in dams and female offspring were reported, respectively. In addition, PFOA-exposed female pups showed stunted mammary gland epithelial branching and growth [39]. However, the hypothesis that the PPAR-α agonistic potential of PFOA was involved was not supported since PPAR-α null mice exhibited normal mammary gland development [39]. In contrast, over expression of PPAR-α during pregnancy in mice impair normal differentiation of the mammary gland [33]. Thus far, data suggest that PPAR-α is the most likely target of PFOA and PFOS with the former having highest affinity in both human and mouse isoforms [33]. Further research is needed to elucidate the role of the PPAR pathway on the mammary gland. Moreover, comparing animal (rodent) data with human health risk must always be done with some caution because the exposure profile and often the body elimination is very different: in animals the exposure is often with single compounds, higher concentration but relatively short time, whereas in human the exposure is lifelong with low doses and complex mixtures.

We found that the serum level of both PFAAs, PFOS and PFOA, as well as the legacy POPs were significant higher in the BC cases compared to the controls but the adjusted OR only indicated significant BC risk for PFOS and PFSA and upon pooling all the chemical groups. However, we found in both cases and controls a high inter-correlation between serum PFSA and PFCA (r = 0.85-0.96, p <0.05) and between these two PFC groups and the legacy POPs (r = 0.42-0.55, p <0.05). Taken that into consideration it cannot be predicted which single compounds or chemical group that play the main role in the observed increase in BC risk.

Preliminary data suggest that some PFCs might be weak xenoestrogens [41]. In adult rats it was found that PFOA led to a decrease in serum testosterone and increase in serum estradiol levels [33]. Future studies are needed to understand the possible roles of PFCs as endocrine disrupters on sex hormone homoeostasis and function.

In our study the total combined serum POP related xenobiotic bio-activities reflects the lipophilic legacy POPs only and does not include the more hydrophilic PFCs. In cases we found that the POP related serum agonistic xenoandrogenic transactivity was significantly associated with the risk of BC as well as a higher number of subjects with significant increased serum agonistic xenoestrogenic transactivity.

With reference to the induced XAR a significant association to BC risk was found both before and after adjustment for related confounders. Using a similar analytical system the biological effects of the total combined xenoestrogens have been set in relation to the risk of BC by N. Olea and his co-workers [67, 68]. It is well established that exposure to estrogens can increase the risk of breast cancer mainly attributed via their potential to increase the receptor induced cell proliferation and thereby expand possible gene damage and mutations in e.g. either tumour suppressing genes or proto-oncogenes causing inactivation or activation of the gene expressions of growth factors that act on mammary epithelial cells in an autocrine or paracrine loop [69]. Thus the body burden of exogenous estrogens such as POP related xenostrogens may further increase the BC risk.

Adams, J.B. [62] proposed a role of adrenal androgens (AA) in the aetiology of breast cancer. Premenopausal women that develop BC tend to have subnormal serum levels of AA and thus lower androgens acting via the AR opposing the estrogenic stimulated cell growth during this life period. In contrast, subjects developing the disease postmenopausal have supranormal levels of AA, and the elevated AA levels stimulate cell growth by the action of 5-androsten-3β, 17β-diol, termed hermaphrodiol, via its combination with the ER in a milieu having low concentrations of the classical E2. It might be hypothesized that the significantly higher combined serum POP related xenoAR activity level in cases has contributed to the BC development predominantly in the postmenopausal cases whereas the higher frequency of cases eliciting significant agonistic POP related xenoestrogenic transactivity might affect the BC risk in both pre- and postmenopausal cases. However, further research is needed before any conclusion can be taken.

In the present study we observed a significant lower level of AhR-TEQ in BC cases compared to their controls although the significance disappeared upon adjustment for the corresponding confounders (age, BMI, pregnancy, breastfeeding and cotinine). The higher serum sum POP/sumPCB levels in cases may explain the observed difference between cases and controls. Non-dioxin-like PCBs are shown to have the potential to antagonize the AhR pathway [70, 71], supporting our unpublished in vitro studies (manuscript in preparation). Moreover, we found in an earlier study that Inuit with high serum levels of PCB had lower AhR-TEQ compared to Europeans with lower PCB levels [55]. Dioxins are reported to have antiestrogenic potentials [72–74]. It can be speculated whether the lower dioxin-like AhR-TEQ level in cases could play a role in BC risk via its lower antiestrogenic potential. Alternatively, it may be explained by differences between cases and controls regarding metabolic pathways involved in the biotransformation of both mono-ortho PCBs and estrogens as suggested by Demers et al. 2002 [75].

In spite of the POPs the traditional diet might also offer some protection against breast cancer because the marine diet is rich in 22:6, n-3 fatty acids and the antioxidant selenium (Se) [76], both factors suggested to have inhibitory effect on breast cancer [77, 78] and chemically induced carcinogenesis in animals [79]. In addition, the intake of n-3 fatty acids is associated to the level of Vitamin A and D. An inverse relation between Zn and breast cancer was reported [80], and solar radiation and thus Vitamin D was suggested to reduce BC risk [81]. The intake of Zn has decreased since 1976 [48, 82]. Thus, by changing to the western-lifestyle food intake the Greenlandic Inuit have a decreased intake of food factors, which might be protective against the risk of BC, including n-3 fatty acids, Zn, Vitamin D and Se.

In this study we found a trend to lower n-3/n-6 fatty acid ratio and lower Se but significantly higher levels of serum Zn in BC cases. To evaluate whether these factors contributed to the development of BC needs further research. However, a recent physiologically based pharmacokinetic (PBPK) model accounting for any given physiologic lifetime history using data on pregnancies, height, weight, and age, estimated the values of physiologic parameters (e.g., organ volume, composition, and blood flow) throughout a woman's entire life was developed. This PBPK model showed the limitations of using a single sample value obtained around the time of diagnosis for lifetime exposure assessment and point out the need to estimate the past POP exposure during time windows that are hypothesized to be mechanistically critical in carcinogenesis [83]. These factors may to some degree explain the controversial reports on POP exposure and breast cancer risk.

In the present study the metals were measured in the whole blood which primarily reflects the actual metal exposure. The limitation in reflecting the body burden prior to the disease will be further discussed in a manuscript particularly focusing on the effects of metals on the risk of breast cancer from the same study population (manuscript is in preparation).

There are some weaknesses in the presented study. Firstly, the few subjects involved, 31 cases and 115 controls, gives a poor statistically power. However, the highly related serum PFC levels with the risk of BC cancer did persist in all our effort to make up a better case-control frequency match. The number of controls in the West region (especially in Nuuk) was higher than other regions and may have impact on the final results. Therefore we also evaluated the data by reducing the number of Nuuk control subjects by matching the age and BMI with the cases in Nuuk to obtain similar cases/control ratio as for the other regions. Similar data was found between the data presented and upon including the reduced and matched data of Nuuk controls (not shown). However, a trend of higher level of organochlorine pesticides (OCPs) and competitive xenoestrogenic transactivity (XERcomp) for cases was observed being significant by including the reduced and matched data of Nuuk controls, whereas the observed significant difference between cases and controls at the highest PCB quartile and XAR transactivity disappeared probably due to reduced statistical power (data not shown).

The presented data show a higher level of PFCs and legacy POP in BC cases indicating that the level of serum POPs in particularly PFCs might be risk factors in the development of breast cancer in Greenlandic Inuit. Furthermore, our data suggest that the higher level of legacy POP related xenoestrogenic and xenoandrogenic agonistic activities in cases compared to controls can contribute to the development of BC in Inuit. Further investigations are needed to document these study conclusions.

The authors declare that they have no competing interests.