Environmental pollutants mimicking the effects of estrogen are suggested to contribute to the high incidence of hormone-related cancers, but supporting data are sparse. A potent estrogen-like activity of the pollutant cadmium, mediated via the estrogen receptor-α, has been shown in vivo. We prospectively examined the association between cadmium exposure and incidence of postmenopausal endometrial cancer. The Swedish Mammography Cohort is a population-based prospective cohort of 30,210 postmenopausal women free of cancer diagnose at baseline (1987) and who completed a food frequency questionnaire at baseline and in 1997. We estimated the dietary cadmium intake based on the questionnaire data and the cadmium content in all foods. During 16.0 years (484,274 person-years) of follow-up between the baseline and mid-2006, we ascertained 378 incident cases of endometrioid adenocarcinoma. The average estimated dietary cadmium intake was 15 μg/day (80% from cereals and vegetables). Cadmium intake was statistically significantly associated with increased risk of endometrial cancer in all women; the multivariate relative risk (RR) was 1.39 [95% confidence interval (CI), 1.04–1.86; Ptrend = 0.019], comparing highest tertile versus lowest. Among never-smoking women with body mass index (BMI) of <27 kg/m2, the RR was 1.86 (95% CI, 1.13–3.08; Ptrend = 0.009). We observed a 2.9-fold increased risk (95% CI, 1.05–7.79) associated with long-term cadmium intake consistently above the median at both baseline 1987 and in 1997 in never-smoking women with low bioavailable estrogen (BMI of <27 kg/m2 and nonusers of postmenopausal hormones). Our results support the hypothesis that cadmium may exert estrogenic effects and thereby increase the risk of hormone-related cancers. [Cancer Res 2008;68(15):6435–41]
- dietary intake
- endometrial cancer
- endocrine disruptor
It has been suggested that environmental pollutants that mimic the effects of estrogen contribute to the high incidence of hormone-related cancers in Western populations ( 1). Although this hypothesis has received extensive media attention, data have hitherto been sparse in supporting such associations in humans ( 2). More recently, cadmium, one of the most serious environmental pollutants ( 1, 3), was proposed as a potent metallestrogen ( 1, 4– 6). Cadmium increases estrogen receptor α–mediated cell proliferation ( 7) and possessed mutagenic properties ( 8, 9). Although some in vitro estrogenicity assays showed no activity ( 10), environmentally relevant doses of cadmium induced several well-characterized estrogenic responses in vivo, including increased uterine weight, hyperplasia, and hypertrophy of the endometrial lining, induction of uterine progesterone receptor, and complement C3 gene expression in animals ( 4). In addition, after in utero exposure, cadmium affected mammary gland development and early onset of puberty in female offspring—both prototypical endocrine disruptor–like responses ( 4). Cadmium has been widely dispersed into the environment through industrial emission, waste incineration, and combustion of fossil fuels. Even in industrially nonpolluted areas, farm land may become contaminated by airborne deposition and by use of cadmium-containing fertilizers and sewage sludge. A high accumulation in agricultural crops results in the fact that plant foods contribute to >80% of the total cadmium intake, with bread, cereals, potatoes, roots, and vegetables being the major sources ( 11).
Endometrial cancer is the cancer most suited for exploring potential estrogenic effects of cadmium. Estrogen unopposed by progesterone or synthetic progestins is the main factor influencing the risk of this cancer ( 12).
We examined prospectively whether the estrogenic properties of cadmium could be confirmed by a direct association between food-cadmium and increased risk of endometrial cancer in the large population-based Swedish Mammography Cohort.
Subjects and Methods
Study population. The Swedish Mammography Cohort was established between 1987 and 1990, when all women who were born between 1914 and 1948 and residing in two counties in Central Sweden received a mailed questionnaire concerning diet, weight, height, education, and parity; 66,651 women responded to the questionnaire (74% response rate). Supplementary information on age at menarche and menopause and postmenopausal hormone use was obtained in part of the cohort. A follow-up questionnaire was sent to all 56,030 participants who were still alive and living in the study area in the autumn of 1997 to update dietary data and to collect information on other life-style factors, including smoking habits, physical activity, age at menarche and menopause, use of postmenopausal hormones, and medical history of diabetes mellitus (additional diabetes information was obtained by computerized linkage of the cohort with the National Hospital Discharge Registry). The response rate was 70%. The study was approved by the Regional Ethical Review Board in Stockholm, Sweden.
Assessment of diet and covariates. At baseline, a 67-item food frequency questionnaire (FFQ) was used to assess dietary intake. An expanded 96-item FFQ was used to update information on dietary intake in 1997. The baseline FFQ has been validated in a subsample of 129 women randomly chosen from the study population. The Pearson correlation coefficients (r) between the FFQ and the mean of 4 1-wk weighted diet records were between 0.5 and 0.8 for the main cadmium-containing foods. In a study of 57 Swedish women, dietary cadmium intake correlated well with tissue cadmium accumulation as assessed by the urinary excretion of the metal; r = 0.7. 1
We created a food-cadmium database based on the cadmium content in all foods available on the Swedish market. The data were provided by the National Food Administration ( 13– 15). We used the average cadmium concentration in each food item because there is no known industrial cadmium contamination of agricultural soil in the study area, or detected geographic variation of cadmium content in foods across Sweden, and most foods are distributed throughout the country by few wholesale companies ( 14), Thus, our estimates of the food-cadmium intake are expected to give a fair representation of the exposure in our study. For only a few food items, Finnish and Danish data were used ( 16– 18). The exposure from air (<1% of total exposure; ref. 19) and drinking water (0.2% of total cadmium intake; ref. 11) is low and was ignored.
For calculation of food, energy, and cadmium intake, we used the age-specific portion sizes based on 5,922 weighed food records kept by 213 randomly selected women from the study area. We adjusted cadmium intake for total energy intake of 1,700 kcal (mean in the cohort) using the residual-regression method ( 20).
We defined menopausal status and age at menopause based on reported age at cessation of menstruation. In case of missing data, women were considered to be postmenopausal if they had bilateral oophorectomy, obtained by computerized linkage of the study population with the National Hospital Discharge Registry, or were ages 55 y or older (because ∼90% of women in the cohort stopped menstruating before the age of 55 y). Body mass index (BMI) was categorized into below or above 27 kg/m2. The validity of BMI based on self-reported weight and height in the Swedish population is high compared with measurements (r = 0.9; ref. 21). We categorized smoking status into “never,” “former,” or “current smoking.” The use of postmenopausal hormones was classified as “never” or “ever use” (we have no information on the type of preparation or the duration of hormone use). We also obtained information on level of education, parity, age at menarche, and age at menopause. We assessed leisure time activity using a validated questionnaire with five predefined categories for time spent per day watching TV/sitting. Inactivity was defined as sitting ≥5 h per day ( 22).
Ascertainment of endometrial cancer. Incident cases of endometrial cancer (International Classification of Disease-10:code: C541) of the histopathologic type: endometrioid adenocarcinoma (Systematized Nomenclature of Medicine code: 81403) were ascertained from baseline (1987) through June 30, 2006, by computerized linkage of the study population to the national and regional Swedish Cancer Registries. The Swedish Cancer Registries provide close to 100% complete case coverage ( 23). By linkage to the Swedish Death and Population Registries, we obtained dates of death and migration from the study area when applicable.
Analytic cohort. We excluded from the baseline population women with incorrect or missing national registration number, those who reported implausible values for energy intake (± 3 SD of mean log-transformed energy), and those with a previous cancer diagnosis or hysterectomy before the baseline. Because diabetes may increase the risk of endometrial cancer ( 24) and that the dietary advice for diabetics involves high consumption of foods high in cadmium, we excluded 2,798 women with diabetes mellitus from the cohort. Because a potential estrogenic effect of cadmium could be masked by the effect of endogenously produced estrogens from the ovaries, we restricted our analyses to women who were postmenopausal at baseline; thus, after exclusion of 28,223 premenopausal and perimenopausal women, the analytic cohort for the primary analyses consisted of 30,210 postmenopausal women followed-up from 1987 to mid-2006.
In the analysis of long-term cadmium intake, we included women who responded to the 1997 questionnaire. In this analysis, women with cancer diagnosis, those having had a hysterectomy, diabetes mellitus, and those who were premenopausal or perimenopausal before 1997 were excluded. As more women had become postmenopausal in 1997, the analysis of long-term cadmium intake consisted of 31,616 postmenopausal women, followed-up from 1997 to mid-2006 (206 ascertained cases of endometrial cancer).
Statistical analysis. Women were categorized into tertiles of cadmium intake at baseline. We estimated relative risks (RR) and 95% confidence intervals (CI) using Cox proportional hazards regression models with attained age (1-y units) as the timescale (Intercooled STATA software version 8.2; StataCorp, LP). Women were censored at date of endometrial cancer diagnosis, hysterectomy, death, or end of follow-up, whichever occurred first. The Schoenfeld's residual test indicated no violation of the proportional hazard assumption. In the first multivariate analysis, we adjusted for attained age, postsecondary education, BMI, use of postmenopausal hormones, smoking status, parity, age at menarche, age at menopause, and leisure time physical inactivity. In additional analyses, we further adjusted for the intake of vegetables, whole grains, and potatoes to examine whether anticarcinogenic substances in these foods are attenuating the observed association between cadmium intake and endometrial cancer risk. Linear trend across categories were tested using the median cadmium values within categories as a continuous variable. Long-term cadmium intake was estimated based on cadmium intake at baseline 1987 and in 1997, and categorized into three categories: consistently below the median, no consistent, and consistently above the median cadmium intake. All reported P values were from two-sided statistical tests.
During a mean of 16.0 years (484,274 person-years) of follow-up between the baseline 1987 and mid-2006, we ascertained 378 incident cases of endometrioid adenocarcinoma among the 30,210 postmenopausal women. The average estimated energy-adjusted cadmium intake in the cohort was 15 μg/d. Women in the highest tertile of cadmium intake were more likely to be never-smokers and to have a postsecondary education, and had a 2-fold higher consumption of vegetables and whole grains compared with those in the lowest tertile ( Table 1 ).
There was a positive dose-response association between cadmium intake and risk of endometrial cancer among all postmenopausal women ( Table 2 ). The highest tertile of cadmium intake was associated with statistically significant multivariate 39% increased risk after adjusting the model for consumption of the major cadmium-contributing foods i.e., vegetables, whole grains, and potatoes.
To reduce the influence of endogenous estrogen exposure, we performed analyses stratified by BMI. In women with BMI of <27 kg/m2, representing a group with lower adipose tissue–derived estrogen exposure, we observed a multivariate-adjusted dose-dependent 52% increased risk (Ptrend = 0.039) between food-derived cadmium and risk of endometrial cancer, comparing the highest tertile of dietary cadmium intake with the lowest ( Table 2). There was no association between cadmium intake and endometrial cancer risk among overweight and obese women.
To take into account a possible effect of smoking on our risk estimates—smokers are additionally exposed to cadmium ( 25) and smoking is associated with decreased risk of endometrial cancer ( 26)—we stratified the analyses by smoking status. In never-smokers, we found a positive dose-response association between dietary cadmium and endometrial cancer risk ( Table 2). The highest tertile of cadmium intake was associated with a multivariate-adjusted 43%, nonstatistically significant, increased risk of the disease compared with the lowest tertile of exposure (Ptrend = 0.055). After further adjustment for major cadmium sources, the RR was 1.56 (95% CI, 1.00–2.45; Ptrend = 0.047). Because information on smoking was only available among two-thirds of the cohort members at baseline, we performed a sensitivity analysis by investigating the association between cadmium and endometrial cancer among never-smokers in the subcohort with follow-up from 1997 to mid-2006, with complete information on smoking. Among never-smokers in this subcohort (n = 16,926 and 127 cancer cases), the highest tertile of cadmium intake was associated with multivariate-adjusted RR of 1.69 (95% CI, 1.05–2.72) compared with the lowest tertile, supporting the accuracy of our assumption to stratify by smoking status at baseline.
To further reduce the influence of estrogens from other sources than cadmium, we stratified the analysis by exogenous estrogen exposure. Stratification by postmenopausal hormone alone had no major effect on the association between food cadmium and endometrial cancer risk.
To further minimize the effect of exposure to both endogenously produced adipose tissue estrogen and to smoking-derived cadmium, we performed a separate analysis among women with BMI <27 kg/m2 and were never-smokers at baseline. We observed a statistically significant positive association with 86% increased risk for the highest versus the lowest tertile of cadmium intake (Ptrend = 0.009; Table 3 ). Further adjustment of the multivariate model for major food sources of cadmium had no effect on the association (data not shown). After further exclusion of women who had used postmenopausal hormones, the corresponding RR was 2.42 (Ptrend = 0.016).
To examine the association between long-term cadmium intake and endometrial cancer, we performed analyses in the subcohort of women with available information on cadmium intake at baseline (1987) and in 1997, with follow-up from 1997 to mid-2006. We observed that never-smoking women, with BMIs of <27 kg/m2 and with long-term consistently higher cadmium intake (corresponding to above the median in the cohort at both baseline 1987 and 10 years later in 1997) had an multivariate-adjusted RR of 2.56 (95% CI, 1.34–4.89) compared with those with cadmium intake below the median at both occasions ( Fig. 1 ). The corresponding RR for never-smoking women with BMIs of <27 kg/m2 and who were nonusers of postmenopausal hormones was 2.86 (95% CI, 1.05–7.79).
In this population-based prospective cohort of postmenopausal women, the estimated cadmium intake was similar to that observed in other areas of Europe and the United States with no particular industrial cadmium contamination ( 11, 18, 27– 31). We found statistically significant positive association between cadmium intake from food and the risk of endometrial cancer. The association was further pronounced in women with low endogenous and exogenous estrogen exposure. Specifically, a 2.9-fold increased risk was associated with a long-term consistently increased cadmium intake, observed among lean or slightly overweight never-smoking women who did not use postmenopausal hormones.
The potential estrogenic effect of cadmium can be masked by the effect of endogenously produced estrogens from the ovaries. Therefore, we restricted our analyses to women who were postmenopausal at baseline. We also stratified our analyses by the use of postmenopausal hormones. The effect of cadmium can also be masked by high BMI, as adipose tissue is the main determinant of bioavailable estrogens in postmenopausal women ( 32). Obesity in postmenopausal women leads to increased peripheral production of estrogens, mainly through the aromatization of androstenedione to estrone in adipose cells and by increased production of estrogen precursors by the adrenal glands ( 12). Our results suggest that, in the absence of increased levels of bioavailable estrogens from adipose tissue in women with BMIs of <27 kg/m2, an estrogenic effect of cadmium may be of importance for the risk of endometrial cancer. Among overweight and obese women, already at increased risk of endometrial cancer because of higher levels of endogenously produced estrogen ( 32), cadmium seems not to pose an additionally increased risk.
In smokers, the inhalation of cadmium-containing tobacco smoke is a dominant source of cadmium exposure ( 25). Tobacco smoking may, however, be a confounding factor because smoking in itself has been reported to be associated with decreased risk of endometrial cancer ( 26). The suggested mechanisms behind this association include an antiestrogenic effect of cigarette smoking on circulating estrogen concentrations due to increased metabolic clearance, a reduction in relative body weight, and an earlier age at menopause.
To our knowledge, this is the first study exploring the risk of endometrial cancer in relation to dietary exposure to cadmium. Because estrogens unopposed by progesterone or synthetic progestins influence the risk of endometrial cancer ( 12, 33), this cancer is well-suited for investigation of the estrogen-like activity shown by cadmium in vivo ( 1, 4– 7). Breast cancer is another cancer where estrogen is a risk factor ( 34). Previously, a statistically significant positive association between concentrations of cadmium in urine and increased risk of breast cancer was observed in a case-control study; odds ratio for highest quartile versus the lowest was 2.29 (95% CI, 1.3–4.2; ref. 35).
The hypothesized protective dietary factors for cancer such as vegetables and dietary fiber ( 36, 37) contribute heavily to the cadmium exposure. A lack of a protective effect on breast cancer recurrence and mortality by a diet high in vegetables and dietary fiber was recently reported in a randomized trial among women previously treated for early stage breast cancer ( 38). Likewise, no association was observed between vegetables and risk of breast cancer in pooling analysis of eight observational cohort studies ( 39). The presence of cadmium in vegetables may counteract the anticarcinogenic substances in these foods and, thus, explain the absence of a protective association between vegetables and some hormone-related cancers ( 36, 37).
The strengths of our study include a prospective population-based design and the availability of data on several established risk factors for endometrial cancer. Furthermore, we had repeated detailed data on diet allowing for assessment of association between long-term cadmium intake and endometrial cancer risk. The prospective design prevents recall bias, and the practically complete follow-up of the study population through linkages to various population-based registers minimized the concern that our findings were affected by differential loss to follow-up.
Several potential limitations may be associated with this analysis. First, measurement error in self-reports is inevitable, which can lead to misclassification of cadmium exposure. However, misclassification in our study is most likely nondifferential due to the prospective design, which may lead to attenuation of the true association. In the case of smoking, we consider misclassification of smoking status of women at baseline not very likely because women who reported to be never-smokers in 1997 were also never-smokers in 1987. Yet, the about one-third of our cohort without information on smoking status at baseline might result in a selective follow-up. The results from our sensitivity analysis among never-smokers in the subcohort who hade complete information smoking did, however, support our results obtained from baseline follow-up. The lack of complete information on the use of postmenopausal hormones at baseline and that neither type of preparation nor duration was available may also introduce some degree of misclassification. Despite the adjustment of many known risk factors for endometrial cancer, we cannot completely exclude the possibility of residual or unmeasured confounding. Although a large cohort, the study is limited by the low number of cases and restricted statistical power after stratification, leading to somewhat imprecise estimates. The number of subanalyses may also increase the risk of chance findings. Nevertheless, the fact that the risk estimates for dietary cadmium showed a clear graded increase with less bioavailable estrogen exposure support the consistency of our findings.
In summary, we provide data showing that the intake of food-derived cadmium was associated with increased risk of endometrial cancer in postmenopausal women, especially those who had low levels of bioavailable estrogen. Although our results need to be confirmed both by further experimental and epidemiologic studies, they support the hypothesis that cadmium is a metallestrogen that may increase the risk of hormone-related cancers.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: The Swedish Cancer Society (A. Åkesson) and the Swedish Research Council/Medicine(A. Åkesson), and Longitudinal Studies (A. Wolk). Funding had no role in the design and conduct of the study, collection management, analysis, and interpretation of the data and preparation, review, or approval of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank L. Jorhem and J. Engman and B. Sundström from the Swedish National Food Administration for providing food-cadmium data.
↵1 B. Julin et al., unpublished observations, 2008.
- Received January 30, 2008.
- Revision received April 1, 2008.
- Accepted May 8, 2008.
- ©2008 American Association for Cancer Research.