Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Dietary Intake Estimates and Urinary Cadmium Levels in Danish Postmenopausal Women

  • Caterina Vacchi-Suzzi ,

    Affiliation Department of Preventive Medicine and Program in Public Health, Stony Brook University, Stony Brook, New York, United States of America

  • Kirsten T. Eriksen,

    Affiliation Danish Cancer Society Research Center, Copenhagen, Denmark

  • Keith Levine,

    Affiliation RTI International Trace Inorganics Department, Research Triangle Park, North Carolina, United States of America

  • Jane McElroy,

    Affiliation Family and Community Medicine, University of Missouri, Columbia, Missouri, United States of America

  • Anne Tjønneland,

    Affiliation Danish Cancer Society Research Center, Copenhagen, Denmark

  • Ole Raaschou-Nielsen,

    Affiliations Danish Cancer Society Research Center, Copenhagen, Denmark, Department of Environmental Science, Aarhus University, Aarhus, Denmark

  • James M. Harrington,

    Affiliation RTI International Trace Inorganics Department, Research Triangle Park, North Carolina, United States of America

  • Jaymie R. Meliker

    Affiliation Department of Preventive Medicine and Program in Public Health, Stony Brook University, Stony Brook, New York, United States of America

Dietary Intake Estimates and Urinary Cadmium Levels in Danish Postmenopausal Women

  • Caterina Vacchi-Suzzi, 
  • Kirsten T. Eriksen, 
  • Keith Levine, 
  • Jane McElroy, 
  • Anne Tjønneland, 
  • Ole Raaschou-Nielsen, 
  • James M. Harrington, 
  • Jaymie R. Meliker



Cadmium is a known carcinogen that can disrupt endocrine signalling. Cigarette smoking and food are the most common routes of non-occupational exposure to cadmium. Cadmium accumulates in the kidney and can be measured in urine, making urine cadmium (U-Cd) a biomarker of long-term exposure. However dietary-cadmium (D-Cd) intake estimates are often used as surrogate indicator of cadmium exposure in non-smoking subjects. It is therefore important to investigate the concordance between D-Cd estimates obtained with Food Frequency Questionnaires and U-Cd.


U-Cd levels were compared with estimated dietary-cadmium (D-Cd) intake in 1764 post-menopausal women from the Danish Diet, Cancer and Health cohort. For each participant, a food frequency questionnaire, and measures of cadmium content in standard recipes were used to judge the daily intake of cadmium, normalized by daily caloric intake. Cadmium was measured by ICP-MS in spot urine sampled at baseline and normalized by urinary creatinine. Information on diet, socio-demographics and smoking were self-reported at baseline.


Linear regressions between U-Cd and D-Cd alone revealed minimal but significant positive correlation in never smokers (R2 = 0.0076, β = 1.5% increase per 1 ng Cd kcal-1, p = 0.0085, n = 782), and negative correlation in current smokers (R2 = 0.0184, β = 7.1% decrease per 1 ng Cd kcal-1 change, p = 0.0006, n = 584). In the full study population, most of the variability in U-Cd was explained by smoking status (R2 = 0.2450, n = 1764). A forward selection model revealed that the strongest predictors of U-Cd were age in never smokers (Δ R2 = 0.04), smoking duration in former smokers (Δ R2 = 0.06) and pack-years in current smokers (Δ R2 = 0.07). Food items that contributed to U-Cd were leafy vegetables and soy-based products, but explained very little of the variance in U-Cd.


Dietary-Cd intake estimated from food frequency questionnaires correlates only minimally with U-Cd biomarker, and its use as a Cd exposure indicator may be of limited utility in epidemiologic studies.


Exposure to the heavy metal cadmium (Cd) has been linked to increased cancer risk, and is listed as an IARC group 1 carcinogen [1]. A high level of Cd exposure, typical of smelter workers, has been linked to increased risk of multiple diseases [2], and chronic low-level Cd exposure has been implicated in diseases of the kidneys [3] and bone [4, 5]. Cigarette smoking is a primary source of exposure in the non-occupationally exposed population [6]. Non-smokers are exposed to Cd mainly through foodstuff, especially grains and vegetables that acquire the metal from soil, which can be naturally rich in Cd as well as contaminated from anthropomorphic sources such as manure, fertilizers and pesticides [7]. Gender differences in Cd metabolism and toxicity exist, and it is increasingly evident that women tend to absorb more Cd from the intestinal tract [8, 9]. One possible explanation for this phenomenon is that women typically have lower iron and zinc stores than men, which can increase intestinal Cd absorption [10].

Cd has a long half-life (> 10 years) in the human body and tends to accumulate in the liver and kidneys. Therefore, long-term body burden is reflected by urinary Cd levels [11, 12]. Urinary Cd (U-Cd) has been used to assess cancer hazard risk [2]; evidence for increased risk of lung, pancreas, breast and endometrial cancer, as well as for decreased bone density, has been found in relation to elevated U-Cd [13, 14].

Food Frequency Questionnaires (FFQ) are often used in human cohort studies to estimate exposure to nutrients, foodborne xenobiotics, and heavy metals in study populations. Although a handful of studies have been reported that compare U-Cd with dietary Cd (D-Cd) intake estimates from FFQs, employing the average Cd content of a market basket of food items and standard recipes and preparation methods reported for each country [1518], ours is the largest such study to date. McElroy et al. used a limited set of questions about consumption of known Cd-rich food, such as crustaceans and internal organs (liver and kidneys) [19]. Olsson et al. used a FFQ specifically designed to capture Cd exposure from both food and beverage items (including water)[20]. The FFQ used by Adams et al. from the Women’s Health Initiative included 122 individual foods or food groups, frequency and portion information [15, 21] referring to the previous 3 months. Julin et al. assessed D-Cd using a 67-item FFQ which was validated against a 1-week weighted diet record for intake of cadmium rich foods (whole grain bread, breakfast cereals, potatoes, root crops, cabbage, and spinach) [22]. Quraishi et al. obtained U-Cd concentration and D-Cd estimates from 1,050 women from the Women Health Initiative (WHI), with an average age of 63.4 years [18]. All of these studies reported a weak correlation between FFQ-estimated D-Cd and U-Cd, but only the Quraishi study provided stratified analysis by smoking status, highlighting the need to further investigate the concordance between estimated and measured exposure in different study populations. This is of particular importance since FFQ-based Cd intake estimates have been used to assess hazard risk ratios of health outcomes in the general population [17, 2327].

In this study we test the comparability of D-Cd, as estimated via the FFQ used in the Danish Diet Cancer and Health Cohort [28], and measured U-Cd levels, normalized by creatinine, in a cohort of 1,764 Danish post-menopausal women that included never, former and current smokers.

Materials and Methods

Ethics Statement

The present study was approved by the regional research ethic committee for Copenhagen and Frederiksberg. Written informed consent was obtained from all study participants upon enrolment into the cohort. The present analysis was carried out without contact to the cohort members or their families. Anonymity of participants was retained by strict data management.

Study Population

From December 1, 1993, through May 31, 1997, a total of 57,053 individuals (29,875 women and 27,178 men), who were aged 50–65 years, born in Denmark, and had no previous cancer diagnosis, were enrolled in the prospective Diet, Cancer and Health cohort [29]. At enrolment, each participant gave a sample of urine and completed a self-administered, interviewer-checked 192 item semi-quantitative FFQ and a questionnaire covering lifestyle habits, including information on smoking history, reproductive history, health status, and social factors. Smoking intensity corresponded to the reported numbers of cigarettes smoked daily, while pack-years was calculated as the number of 20-cigarette packs smoked daily, multiplied by the years of smoking. Exposure to second-hand smoke was reported as a categorical variable to reflect exposure to cigarette smoke either in the household or at work, starting from age 30 and above. In total 56,999 persons filled in the detailed dietary questionnaires. Of these, we obtained U-Cd levels from 1,764 postmenopausal women who were selected to be part of a case-cohort study on Cd and breast cancer [24]; 896 women developed breast cancer from 4 years after baseline visit and through 2012, and 868 women did not develop breast cancer between baseline and 2012. Case status will be considered in the analyses presented.

Dietary Cadmium and Nutrient Intake Estimates

Daily dietary intake of calories (kcal day-1), cadmium (D-Cd, ngn day-1), zinc (Zn) and iron (Fe) (mg day-1) were obtained for each participant based on the 192 item semi-quantitative FFQ filled out at enrolment, as previously described [24, 28, 3033]. Dietary intake estimates of Zn and Fe were obtained by summing estimates from food and supplement consumption. Since D-Cd intake estimates and the daily caloric intake were highly correlated, D-Cd estimates were normalized bycaloric intake (ng kcal-1), according to the nutrient density approach described by Willet et al. [34].n, according to the nutrient density approach described by Willet et al. The Danish Food Monitoring Programme for Nutrients and Contaminants, 1993–97, along with the FFQ, were used to estimate individual daily cadmium intake [35]. The 5-year monitoring period 1993–97 was used, since it matches with the enrolment period of the DCH cohort. Estimates of Cd, Fe and Zn content were obtained for over 80 food items commonly available on the Danish market.

Urinary Cadmium Levels

Cd was measured inurine samples, which wereere collected in transparent polypropylene cups (USON Plast, Denmark) and stored in transparent 1 ml polypropylene cryotubes (NUNC, Denmark). The urine was never in contact with any metallic equipment, and the suppliers guaranteed that Cd was not used in the manufacturing of collection and storage materials. Urinary cadmium concentrations were determined using a Thermo X-Series 2 (Bremen, Germany) inductively coupled plasma mass spectrometer (ICP-MS) following digestion in the presence of high-purity acids and oxidants in a Class 100 clean hood to prevent contamination by atmospheric particulates. Urinary creatinine concentrations were quantified using a Cayman Chemicals Creatinine Assay Kit No. 500701 (Cayman Chemicals, Ann Arbor, MI, USA) with UV-VIS measurement at 500 nm employing a Beckman Coulter DU800 UV/VIS Spectrometer (Beckman Instruments Inc., Brea, CA, USA). Samples below the limit of detection (LOD) (N = 23) were assigned to the batch Cd LOD √2−1[36, 37]. For each urine sample, Cd concentration (μg L-1) was divided by the determined creatinine amount (g L-1) to obtain creatinine-adjusted urinary cadmium levels (U-Cd) in μg Cd g creatinine-1. LOD varied between <0.0011 and 0.047 μg/L across 30 discrete analytical batches. Average recovery of NIST standard reference material (SRM) 2668was 90.2%, across all analysis batches. The lower and higher levels of Cd in the NIST-SRM 2668 were 1.06 ± 0.05 and 16.40 ± 0.25 μg/L. Random incurred sample re-testing of 5% of the samples showed Pearson correlation r = 0.90 with initial measurements.

Statistical Analyses

Spearman’s r coefficient and p value were calculated for creatinine-adjusted U-Cd and D-Cd. The ANOVA test wasas used to identify differences in the mean value of Cd measures (either D-Cd estimates or log-transformed creatinine-adjusted U-Cd) across categories of 10 covariates (creatinine-adjusted U-Cd, D-Cd, smoking status, pack-years, caseca status, energy intake, BMI, age, iron intake and parity).

Linear regression models were used to investigate the association between D-Cd, as the independent variable, and creatinine-adjusted U-Cd, as the dependent variable. Models were adjusted for a priori defined potential covariates: age, smoking status (never, former, current), pack-years (packs-per-year and years smoking), and Fe intake (mg Fe day-1). Regression analyses were run for all participants and stratified by smoking status. We used an interaction term to test if the association between D-Cd and U-Cd differed by age, sex, pack-years, daily energy intake, BMI, daily Zn and Fe intake.

An unsupervised forward selection procedure was used to explore the relevance of specific dietary and demographic items as predictors of urinary Cd levels. Briefly, we constructed general linear models with continuous log-transformed creatinine-adjusted U-Cd as a dependent variable for never, former and current smokers and for all participants. A model optimization procedure was used that added variables at each step that provide the greatest value of the adjusted R2 statistic, stopping at the step where the significance level corresponding to the addition of a predictor was greater than 0.2. The predictor variables available to the model selection procedure were: red meat (g day-1), soy (g day-1), all fish and seafood (g day-1), fruit (g day-1), all meat (g day-1), potatoes (g day-1), vegetables and fruit (g day-1), fruits and grains (g day-1), all grains (g day-1), leafy vegetables (g day-1), mushrooms (g day-1), whole grains (g day-1), white grains (g day-1), tea (g day-1), wine (g day-1), beer (g day-1), spirits (g day-1), Fe (mg day-1) and Zn (mg day-1), age (years), BMI (kg m-2), smoking duration (years, for former smokers only), number of children (0 (reference), 1–2, ≥3), D-Cd (μg day-1), D-Cd normalized by body weight (μg day-1 kg-1), D-Cd normalized by daily caloric intake (ng kcal-1), smoking status (never, former, current), later development of breast cancer (no, yes), pack-years (n. of packs day-1 years of smoking, for current smokers model only), exposure to second-hand cigarette smoke at work or home since 50 years of age (yes or no, for never smokers only), number of decades of exposure to second-hand cigarette smoke at work or home since 40 years of age (0 to 4, for never smokers only) and average daily caloric intake (kcal day-1). The food items included in the selection procedure were chosen based on known literature [24, 38, 39] and information contained in the ATSDR Cadmium Toxic Substance Portal [40]. We were not able to consider second-hand smoke among smokers because of collinearity with smoking status. In order to exclude the possibility that the selected variables would be selected because of association with creatinine, rather than with U-Cd [41], we also used the unsupervised forward selection procedure to construct models using ln(U-Cd) as the dependent variable, and added urinary creatinine (g L-1) as a predictor instead.

Creatinine-adjusted U-Cd was transformed by the natural logarithm to account for non-normal distribution. The β estimates were natural-logarithm back transformed, followed by subtractingsubtracting 1 and multiplyingying by 100 [(eβ-1)*100], obtaining the % change in the dependent variable U-Cd per a 1-unit change in the independent variables listed. The procedure PROC GLMSELECT was used for the model optimization procedure, specifying the option “selection = forward (select = ADJRSQ stop = SL SLE = 0.2)”. All analyses were performed in SAS version 9.3 (SAS Institute, Cary, North Carolina, USA).

For each regression model, we have reported R2 adjusted by degrees of freedom [42], β values, and p values.


Demographic characteristics and descriptive statistics of our study population are summarized in Table 1. U-Cd mean ± SD was found to be 0.46 ± 0.60 μg Cd L-1 without creatinine adjustment and 0.70 ± 0.62 μg Cd g creatinine-1, while D-Cd mean ± SD was 14.00 ± 4.35 (ngn Cd day-1).

As can be seen in Table 2, as U-Cd increased, D-Cd decreased. This inverse relation was confirmed by a negative rank correlation coefficient (Spearman r = -0.14, p = <0.0001) for the entire population. The correlation was positive, but weak and borderline significant, in never smokers (Spearman r = 0.07, p = 0.06). As expected, never smokers exhibited lower levels of creatinine-adjusted U-Cd than both former and current smokers (ANOVA p <0.01, Table 2), and U-Cd levels were higher in current smokers than former smokers and correlated toto increasing pack-years (ANOVA p <0.01). Conversely, D-Cd was lower in current smokers than in never smokers. Higher levels of U-Cd were found in older women and also increased with children parity (ANOVA p <0.01, Table 2). U-Cd was marginally lower in individuals with higher Fe intake while D-Cd was higher in those with increased Fe intake and lower in those with greater energy intake. Participants that reported exposure to second-hand smoke had lower U-Cd and D-Cd values. There was no difference in U-Cd or D-Cd levels observed between cancer cases and controls, or BMI ranges.

Table 2. Mean ± Standard Deviation (SD) of U-Cd (μg Cd g creatinine-1) and D-Cd (ng Cd kcal-1) across cohort characteristics.

The non-adjusted association between creatinine-adjusted U-Cd and D-Cd in a regression analysis was small (R2 = 0.007), negative (β = -4.79% change in U-Cd per unit increase of D-Cd), and significant (p < 0.01) in the whole study population (Table 3). However the association disappeared (p = 0.67) when the model was controlled for age, smoking status, and Fe intake, with smoking status explaining the largest proportion of the variance. In adjusted models, the association between creatinine-adjusted U-Cd and D-Cd was positive in never smokers (β = 5.13% change in U-Cd per unitunit increase of D-Cd, p < 0.01) and negative in current smokers (β = -5.08% change in U-Cd per unit increase of D-Cd, p = 0.01) (Table 3). However, the overall R2 values in these cases remained small. D-Cd was not associated with U-Cd in former smokers. D-Cd was not associated with U-Cd in adjusted models when the study population was stratified into tertiles of age or Fe intake, or by disease status, and there was no evidence of interaction between D-Cd and age, Fe, or disease (data not shown).

Table 3. Associations between D-Cd (ng kcal-1) and creatinine-adj. U-Cdd (μg Cd g creatinine-1).

A forward model selection procedure, aimed to identify factors predictive of U-Cd, selected different combinations of 17 out of 36 possible predictor variables across 4 different models (Table 4). The maximum R2 obtained in the entire population was 0.22, and was achieved with a combination of 9 variables that included (in order of decreasing R2 magnitude) being current or former smoker, age, consuming tea, soy, or wine, having 3 or more children, consuming red meat or vegetables/fruit products. In never smokers, the optimal fit included 10 predictor variables (age, vegetables/fruit, red meat, soy, wine, fruit, fish, whole grains, having 3 or more children and spirits), resulting in an R2 of 0.09. In former smokers, the model selection procedure produced an optimal R2 of 0.09 using a combination of smoking duration, wine, having 3 or more children, soy, age and Zn. In current smokers, a maximum R2 of 0.10 was obtained from a model including a combination of 7 predictor variables, including pack-years, tea, soy, BMI, age, D-Cd and energy intake. The variables that generally contributed the most to the R2 in the different models were smoking status (for the general population), age (for never smokers), smoking duration (for former smokers), and pack-years (for current smokers). A higher number of childbirths was consistently associated with increasing levels of U-Cd. In terms of dietary contributors, soy product consumption was consistently associated with higher U-Cd with the largest estimates across all models (β ≥0.20% change in U-Cd per each g day-1 increase of soy consumption). Wine was negatively associated with U-Cd (β max. -0.06% decrease in U-Cd per each g day-1 increase of wine consumption). No other food variable was consistently associated with U-Cd in all models. Consumption of red meat, wine and tea appeared consistently inversely associated with U-Cd, although their contributions were generally small and not always significant (β < -0.3% change in U-Cd per unit increase of either item consumption). Dietary Zn was marginally inversely correlated with U-Cd in former smokers (β = -0.76% change in U-Cd per mg Zn day-1); dietary Fe was not selected in any models as a significant contributor to U-Cd levels. When we used non-normalized ln(U-Cd) as dependent variable, and added urinary creatinine to the predictors, we obtained very similar results in all selection models, with the main difference that the best predictor of U-Cd levels was creatinine itself.

Table 4. Forward model selection to predict creatinine-adj. U-Cda (μg Cd g creatinine-1).

Parameter estimates are listed for the factors selected by the procedure for inclusion in each model (p value cut-off = 0.2).

The overall concordance across quartiles was 26%, not dissimilar from what would be obtained by chance (S1 Table).


We report a weak association between measured U-Cd and dietary estimates of Cd in this cohort of 1,764 post-menopausal Danish women. This association was found to be positive among never smokers and negative among current smokers in adjusted regression models. Our U-Cd and D-Cd results were in line with levels previously measured in non-occupationally exposed populations and were below the internationally established safety thresholds [11, 13, 40].

Since smoking is a major source of cadmium exposure, we hypothesized that never smokers are most likely to show the strongest association between D-Cd and U-Cd. Our results confirmed our hypothesis and showed a significant positive association between D-Cd and U-Cd in never smokers. The adjusted R2, however, was small, suggesting that overall the intake estimates are of limited use to approximate cumulated Cd uptake, even in never smokers. In contrast, we found a small, significant, and negative, association between D-Cd and U-Cd in current smokers. This could be explained by lower consumption of foods rich in cadmium, such as grains and vegetables, among smokers [43, 44]. Consistent with the current understanding that U-Cd reflects long-term Cd storage in the kidneys [8, 15, 19], yet in contrast to reports by Chaumont et al., 2013 and Paschal et al., 2000 [45, 46], we found significantly higher levels of U-Cd in former smokers than in never smokers. We also found that age is a strong predictor of U-Cd in never smokers, consistent with previous reports [9, 16, 45], which is especially striking since our cohort included only postmenopausal women within a limited age range. Second-hand smoke was not associated with U-Cd among never smokers, suggesting that passive smoke was not a sizable contributor to U-Cd levels in our population.

Childbirth was positively associated with U-Cd levels, which can be explained by depleted Fe stores during pregnancy [47]. Lower body stores of divalent cations (specifically Fe and Zn) have been implicated in the absorption and toxicity of Cd [48]. However our variables for Fe or Zn intake were not consistently selected as U-Cd determinants during the selection procedure across models (Table 4). This observation may be explained by the fact that after menopause, Fe body stores increase [49] or simply by the higher uncertainty associated with estimating Fe and Zn levels from dietary sources, rather than measuring blood Zn and Fe content [22].

We found that soy was the most significant dietary predictor of U-Cd, consistent with a previous finding [15], even though the consumption of soy in our cohort was quite low (on average less than 1g day-1). Consumption of tea and red meat were generally associated with lower levels of U-Cd, although only marginally. An inverse relation between Cd in blood and total meat consumption was also reported by Bjermo et al. [48]. An inverse correlation was found with wine as well, andand while the percent change per gram appears modest, each glass of wine contains approximately 148 g resulting in -8.9% U-Cd per daily glass among former smokers.

The correlation between estimated D-Cd and measured U-Cd in never smokers (r = 0.07) was much smaller than what was reported by measuring Cd directly in four days of food duplicates of never smokers (r = 0.38–0.43) and correlating it with U-Cd in spot urine samples [22, 50]. That report is somewhat surprising since U-Cd is thought to be a marker of long-term cumulative exposure, and therefore one might think that it should not correlate as well with measured Cd from recently collected duplicate food samples. The authors, Julin et al. (2011), argue that the dietary samples collected reflect long-term food preferences [22], which is a possible explanation for the substantial correlation between Cd in the food duplicates and measured U-Cd. Another explanation might be that the influence of acute exposure on U-Cd may not be tiny. While some studies have reported low intra-individual temporal variability in U-Cd (ICC = 0.7–0.9) [51, 52], others suggest it may be larger (ICC~0.5) [9, 53, 54]., others suggest it may be larger (ICC~0.5) The smaller correlation between FFQ-derived D-Cd and measured U-Cd in never smokers (r = 0.07) in this study, therefore, mayay reflect uncertainties in the FFQ-derived Cd intake estimate or intra-individual variability in the U-Cd measure.

While not perfect, FFQs have been validated for estimating energy, macronutrients (protein, fat, sodium and potassium), and some micronutrients [21, 28, 55, 56]. Energy adjustment is essential in FFQ-derived estimates because estimates of nutrients and contaminants are often highly correlated with energy intake [34], as they were here., as they were here. In this study we used the nutrient density approach dividing D-Cd by total calorie intake; a sensitivity analysis using the residual approach produced highly similar results. FFQs are seldom validated against other dietary assessment methods for dietary contaminants, and to our knowledge have not been validated for D-Cd. Regional variability in cadmium levels in foodstuffs also likely contributes to errors in the FFQ-derived D-Cd estimates [57]. In addition, exposure estimates from FFQs are known to be influenced by recall bias and inconsistencies in portion size reporting, especially if it is administered only one time [58, 59]. The results from our current study and previously published studies [1518, 22] suggest that D-Cd from an FFQ has limited value as a predictor of U-Cd.

Recent epidemiologic studies have reported on the association between D-Cd from FFQs and several health outcomes [17, 24, 25, 27, 6062]. The majority of the studies have reported null findings, and have suggested that one of the reasons may be non-differential exposure misclassification which tends to attenuate results toward the null [59], resulting from the FFQ-derived D-Cd estimate. While our results cannot confirm this error in the FFQ-derived D-Cd estimate, we clearly show a lack of association with U-Cd, a good biomarker of long-term exposure [11, 12], suggesting that non-differential exposure misclassification may be contributing to the epidemiologic findings. On the other hand, it is possible that while U-Cd is strongly correlated with Cd stored in the kidneys [63], in whichCd has an estimated half-life of ~45 years [64], it, it may not reflect doses of relevance in other organs.

The design and performance of this study has multiple strengths. First, it is the largest study investigating concordance between U-Cd and FFQ-estimated dietary Cd exposure. It relies on a validated FFQ of over 192 food items and recipes [28]; this thorough study design allowed us to investigate the association of Fe and Zn intake with U-Cd. There are large numbers of women among different categories of smoking status, to allow us to account for the role of smoking on the association between D-Cd and U-Cd. The trace inorganics laboratory used a robust analytical method with good standard reference material recoveries, a very low detection limit for U-Cd, and displayedgood reproducibility in the U-Cd measure. This study which enrolled a large sample of Danish citizens allowed for applying dietary Cd estimates from food basket survey and typical recipes to a relatively homogeneous population.

The limitations of this study must be carefully considered when interpreting the results. The Danish Diet Cancer and Health subcohort utilized in this study was not a fully representative sample of Denmark women, given the limited age range, and selection of only post-menopausal women [24]. Additionally, approximately 50% of this study population developed breast cancer 4+ years after urine collection and FFQ completion; although results did not differ when cancer cases were excluded from the analysis. Another limitation is the lack of a standardized collection time for the spot urine collection. First-morning urine and 24-h collection are generally considered better samples to estimate U-Cd [52]. Another limitation is the lack of a standardized collection time for the spot urine collection. First-morning urine and 24-h collection are generally considered better samples to estimate U-Cd.

We also did not include information about toxicokinetics in our study. Toxicokinetic models have been developed to approximate Cd body burden based on dietary intake [11, 12], however when duplicate diet D-Cd was entered in such a toxicokinetic model, derived from the Nordberg-Kjellström equation [22, 65], the D-Cd–U-Cd correlation increased only modestly (from r = 0.43 to r = 0.54) despite adding the calculated intestinal absorption rate and mobilization to the model [11]. These results suggest that the duplicate diet component is more important than the toxicokinetic component in correlating D-Cd with U-Cd.


We found only weak association between FFQ-derived dietary intake estimates and urinary levels of cadmium in this large older female population. Estimated dietary cadmium intake from FFQ likely has limited use in epidemiological studies.

Supporting Information

S1 Table. Cross classification of D-Cd and creatinine-adjusted U-Cd in never smokers (n = 782). D-Cd/U-Cd column sum, (column %).



Author Contributions

Conceived and designed the experiments: ORN JM. Performed the experiments: KL JMH. Analyzed the data: CVS KTE. Contributed reagents/materials/analysis tools: AT ORN. Wrote the paper: CVS KTE JM ORN AT KL JRM.


  1. 1. International Agency for Research on Cancer (IARC). IARC Monographs on the Evaluation of Carcinogenic Risks of Humans 2014. Available:
  2. 2. Nawrot T, Plusquin M, Hogervorst J, Roels HA, Celis H, Thijs L, et al. Environmental exposure to cadmium and risk of cancer: a prospective population-based study. The lancet oncology. 2006;7(2):119–26. Epub 2006/02/04. doi: 10.1016/s1470-2045(06)70545-9 pmid:16455475.
  3. 3. Suwazono Y, Sand S, Vahter M, Filipsson AF, Skerfving S, Lidfeldt J, et al. Benchmark dose for cadmium-induced renal effects in humans. Environmental health perspectives. 2006;114(7):1072–6. Epub 2006/07/13. pmid:16835061; PubMed Central PMCID: PMCPmc1513341.
  4. 4. Gallagher CM, Kovach JS, Meliker JR. Urinary cadmium and osteoporosis in U.S. Women >or = 50 years of age: NHANES 1988–1994 and 1999–2004. Environmental health perspectives. 2008;116(10):1338–43. doi: 10.1289/ehp.11452 pmid:18941575; PubMed Central PMCID: PMC2569092.
  5. 5. Schutte R, Nawrot TS, Richart T, Thijs L, Vanderschueren D, Kuznetsova T, et al. Bone resorption and environmental exposure to cadmium in women: a population study. Environmental health perspectives. 2008;116(6):777–83. Epub 2008/06/19. doi: 10.1289/ehp.11167 pmid:18560534; PubMed Central PMCID: PMCPmc2430234.
  6. 6. Marano KM, Naufal ZS, Kathman SJ, Bodnar JA, Borgerding MF, Garner CD, et al. Cadmium exposure and tobacco consumption: Biomarkers and risk assessment. Regulatory toxicology and pharmacology: RTP. 2012;64(2):243–52. doi: 10.1016/j.yrtph.2012.07.008 pmid:22902766.
  7. 7. Cabrera C, Ortega E, Lorenzo ML, Lopez MC. Cadmium contamination of vegetable crops, farmlands, and irrigation waters. Reviews of environmental contamination and toxicology. 1998;154:55–81. Epub 1998/01/01. pmid:9414631.
  8. 8. Adams SV, Newcomb PA. Cadmium blood and urine concentrations as measures of exposure: NHANES 1999–2010. Journal of exposure science & environmental epidemiology. 2014;24(2):163–70. Epub 2013/09/05. doi: 10.1038/jes.2013.55 pmid:24002489; PubMed Central PMCID: PMCPmc4079252.
  9. 9. Gunier RB, Horn-Ross PL, Canchola AJ, Duffy CN, Reynolds P, Hertz A, et al. Determinants and within-person variability of urinary cadmium concentrations among women in northern California. Environmental health perspectives. 2013;121(6):643–9. doi: 10.1289/ehp.1205524 pmid:23552363; PubMed Central PMCID: PMC3672909.
  10. 10. Vahter M, Akesson A, Liden C, Ceccatelli S, Berglund M. Gender differences in the disposition and toxicity of metals. Environmental research. 2007;104(1):85–95. doi: 10.1016/j.envres.2006.08.003 pmid:16996054.
  11. 11. Amzal B, Julin B, Vahter M, Wolk A, Johanson G, Akesson A. Population toxicokinetic modeling of cadmium for health risk assessment. Environmental health perspectives. 2009;117(8):1293–301. doi: 10.1289/ehp.0800317 pmid:19672411; PubMed Central PMCID: PMC2721875.
  12. 12. Nordberg GF, Nogawa K, Nordberg M, Friberg L. Handbook on the Toxicology of Metals: Cadmium. Amsterdam: Elsevier; 2007. p. 445–86.
  13. 13. Garcia-Esquinas E, Pollan M, Tellez-Plaza M, Francesconi KA, Goessler W, Guallar E, et al. Cadmium Exposure and Cancer Mortality in a Prospective Cohort: The Strong Heart Study. Environmental health perspectives. 2014. Epub 2014/02/18. doi: 10.1289/ehp.1306587 pmid:24531129.
  14. 14. McElroy JA, Shafer MM, Trentham-Dietz A, Hampton JM, Newcomb PA. Cadmium exposure and breast cancer risk. Journal of the National Cancer Institute. 2006;98(12):869–73. doi: 10.1093/jnci/djj233 pmid:16788160.
  15. 15. Adams SV, Newcomb PA, Shafer MM, Atkinson C, Bowles EJ, Newton KM, et al. Sources of cadmium exposure among healthy premenopausal women. The Science of the total environment. 2011;409(9):1632–7. Epub 2011/02/22. doi: 10.1016/j.scitotenv.2011.01.037 pmid:21333327; PubMed Central PMCID: PMCPmc3056571.
  16. 16. McElroy JA, Shafer MM, Hampton JM, Newcomb PA. Predictors of urinary cadmium levels in adult females. The Science of the total environment. 2007;382(2–3):214–23. Epub 2007/06/05. doi: 10.1016/j.scitotenv.2007.04.015 pmid:17544058.
  17. 17. Julin B, Wolk A, Bergkvist L, Bottai M, Akesson A. Dietary cadmium exposure and risk of postmenopausal breast cancer: a population-based prospective cohort study. Cancer research. 2012;72(6):1459–66. doi: 10.1158/0008-5472.CAN-11-0735 pmid:22422990.
  18. 18. Quraishi SM, Adams SV, Schafer M, Meliker JR, Li WJ, Carbone L, et al. Cadmium Exposure Measured in Urine and Diet in the Women’s Health Initiative. In press.
  19. 19. McElroy JA, Shafer MM, Trentham-Dietz A, Hampton JM, Newcomb PA. Urinary cadmium levels and tobacco smoke exposure in women age 20–69 years in the United States. Journal of toxicology and environmental health Part A. 2007;70(20):1779–82. Epub 2007/09/22. doi: 10.1080/15287390600754953 pmid:17885936.
  20. 20. Olsson IM, Bensryd I, Lundh T, Ottosson H, Skerfving S, Oskarsson A. Cadmium in blood and urine—impact of sex, age, dietary intake, iron status, and former smoking—association of renal effects. Environmental health perspectives. 2002;110(12):1185–90. pmid:12460796; PubMed Central PMCID: PMC1241104.
  21. 21. Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement Characteristics of the Women’s Health Initiative Food Frequency Questionnaire. Annals of epidemiology. 1999;9(3):178–87. doi: pmid:10192650
  22. 22. Julin B, Vahter M, Amzal B, Wolk A, Berglund M, Akesson A. Relation between dietary cadmium intake and biomarkers of cadmium exposure in premenopausal women accounting for body iron stores. Environmental health: a global access science source. 2011;10:105. doi: 10.1186/1476-069X-10-105 pmid:22177271; PubMed Central PMCID: PMC3265442.
  23. 23. Sawada N, Iwasaki M, Inoue M, Takachi R, Sasazuki S, Yamaji T, et al. Long-term dietary cadmium intake and cancer incidence. Epidemiology (Cambridge, Mass). 2012;23(3):368–76. Epub 2012/03/15. doi: 10.1097/EDE.0b013e31824d063c pmid:22415110.
  24. 24. Eriksen KT, Halkjaer J, Sorensen M, Meliker JR, McElroy JA, Tjonneland A, et al. Dietary cadmium intake and risk of breast, endometrial and ovarian cancer in Danish postmenopausal women: a prospective cohort study. PloS one. 2014;9(6):e100815. Epub 2014/06/26. doi: 10.1371/journal.pone.0100815 pmid:24963789; PubMed Central PMCID: PMCPmc4071014.
  25. 25. Adams SV, Newcomb PA, White E. Dietary cadmium and risk of invasive postmenopausal breast cancer in the VITAL cohort. Cancer causes & control: CCC. 2012;23(6):845–54. doi: 10.1007/s10552-012-9953-6 pmid:22527162; PubMed Central PMCID: PMC3375316.
  26. 26. Akesson A, Julin B, Wolk A. Long-term dietary cadmium intake and postmenopausal endometrial cancer incidence: a population-based prospective cohort study. Cancer research. 2008;68(15):6435–41. doi: 10.1158/0008-5472.CAN-08-0329 pmid:18676869.
  27. 27. Adams SV, Quraishi SM, Shafer MM, Passarelli MN, Freney EP, Chlebowski RT, et al. Dietary cadmium exposure and risk of breast, endometrial, and ovarian cancer in the Women's Health Initiative. Environmental health perspectives. 2014;122(6):594–600. Epub 2014/03/19. doi: 10.1289/ehp.1307054 pmid:24633137; PubMed Central PMCID: PMCPmc4050510.
  28. 28. Overvad K, Tjonneland A, Haraldsdottir J, Ewertz M, Jensen OM. Development of a semiquantitative food frequency questionnaire to assess food, energy and nutrient intake in Denmark. International journal of epidemiology. 1991;20(4):900–5. pmid:1800428.
  29. 29. Tjonneland A, Olsen A, Boll K, Stripp C, Christensen J, Engholm G, et al. Study design, exposure variables, and socioeconomic determinants of participation in Diet, Cancer and Health: a population-based prospective cohort study of 57,053 men and women in Denmark. Scandinavian journal of public health. 2007;35(4):432–41. Epub 2007/09/06. doi: 10.1080/14034940601047986 pmid:17786808.
  30. 30. Larsen EH, Andersen NL, Moller A, Petersen A, Mortensen GK, Petersen J. Monitoring the content and intake of trace elements from food in Denmark. Food additives and contaminants. 2002;19(1):33–46. Epub 2002/01/31. doi: 10.1080/02652030110087447 pmid:11817374.
  31. 31. Haraldsdottir J, Tjonneland A, Overvad K. Validity of individual portion size estimates in a food frequency questionnaire. International journal of epidemiology. 1994;23(4):786–96. pmid:8002194.
  32. 32. Tjonneland A, Haraldsdottir J, Overvad K, Stripp C, Ewertz M, Jensen OM. Influence of individually estimated portion size data on the validity of a semiquantitative food frequency questionnaire. International journal of epidemiology. 1992;21(4):770–7. pmid:1521982.
  33. 33. Tjonneland A, Overvad K, Haraldsdottir J, Bang S, Ewertz M, Jensen OM. Validation of a semiquantitative food frequency questionnaire developed in Denmark. International journal of epidemiology. 1991;20(4):906–12. pmid:1800429.
  34. 34. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. The American journal of clinical nutrition. 1997;65(4 Suppl):1220S–8S; discussion 9S-31S. Epub 1997/04/01. pmid:9094926.
  35. 35. Jørgensen K, Larsen EH, Petersen A, Lund KH, Hilbert G, Andersen NL, et al. Chemical contaminants, Food monitoring, 1993–1997. Part 2.: Ministry of Food, Agriculture and Fisheries, Danish Veterinary and Food Administration; 2001.
  36. 36. Finkelstein MM, Verma DK. Exposure estimation in the presence of nondetectable values: another look. AIHAJ: a journal for the science of occupational and environmental health and safety. 2001;62(2):195–8. Epub 2001/05/03. pmid:11331991.
  37. 37. Hornung RW, Reed LD. Estimation of Average Concentration in the Presence of Nondetectable Values. Applied Occupational and Environmental Hygiene. 1990;5(1):46–51. doi: 10.1080/1047322X.1990.10389587.
  38. 38. Jarup L, Berglund M, Elinder CG, Nordberg G, Vahter M. Health effects of cadmium exposure—a review of the literature and a risk estimate. Scandinavian journal of work, environment & health. 1998;24 Suppl 1:1–51. Epub 1998/05/07. pmid:9569444.
  39. 39. van Wijngaarden E, Singer EA, Palapattu GS. Prostate-specific antigen levels in relation to cadmium exposure and zinc intake: results from the 2001–2002 National Health and Nutrition Examination Survey. The Prostate. 2008;68(2):122–8. doi: 10.1002/pros.20668 pmid:18044729.
  40. 40. ATSDR (Agency for Toxic Substances & Disease registry). Toxicological profile for Cadmium: Atlanta, GA; 2011. Available:
  41. 41. Barr DB, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, Pirkle JL. Urinary creatinine concentrations in the U.S. population: implications for urinary biologic monitoring measurements. Environmental health perspectives. 2005;113(2):192–200. pmid:15687057; PubMed Central PMCID: PMC1277864.
  42. 42. Miles J. R Squared, Adjusted R Squared. Wiley StatsRef: Statistics Reference Online: John Wiley & Sons, Ltd; 2014.
  43. 43. Cade JE, Margetts BM. Relationship between diet and smoking—is the diet of smokers different? Journal of epidemiology and community health. 1991;45(4):270–2. Epub 1991/12/01. pmid:1795144; PubMed Central PMCID: PMCPmc1059458.
  44. 44. Preston AM. Cigarette smoking-nutritional implications. Progress in food & nutrition science. 1991;15(4):183–217. Epub 1991/01/01. pmid:1784736.
  45. 45. Chaumont A, Voisin C, Deumer G, Haufroid V, Annesi-Maesano I, Roels H, et al. Associations of urinary cadmium with age and urinary proteins: further evidence of physiological variations unrelated to metal accumulation and toxicity. Environmental health perspectives. 2013;121(9):1047–53. doi: 10.1289/ehp.1306607 pmid:23774576; PubMed Central PMCID: PMC3764089.
  46. 46. Paschal DC, Burt V, Caudill SP, Gunter EW, Pirkle JL, Sampson EJ, et al. Exposure of the U.S. population aged 6 years and older to cadmium: 1988–1994. Archives of environmental contamination and toxicology. 2000;38(3):377–83. Epub 2000/02/10. pmid:10667937.
  47. 47. Akesson A, Berglund M, Schutz A, Bjellerup P, Bremme K, Vahter M. Cadmium exposure in pregnancy and lactation in relation to iron status. American journal of public health. 2002;92(2):284–7. pmid:11818307; PubMed Central PMCID: PMC1447058.
  48. 48. Bjermo H, Sand S, Nalsen C, Lundh T, Enghardt Barbieri H, Pearson M, et al. Lead, mercury, and cadmium in blood and their relation to diet among Swedish adults. Food and chemical toxicology: an international journal published for the British Industrial Biological Research Association. 2013;57:161–9. Epub 2013/03/30. doi: 10.1016/j.fct.2013.03.024 pmid:23537601.
  49. 49. Vahter M, Berglund M, Akesson A. Toxic metals and the menopause. The journal of the British Menopause Society. 2004;10(2):60–4. Epub 2004/06/23. doi: 10.1258/136218004774202364 pmid:15207026.
  50. 50. Shimbo S, Zhang ZW, Moon CS, Watanabe T, Nakatsuka H, Matsuda-Inoguchi N, et al. Correlation between urine and blood concentrations, and dietary intake of cadmium and lead among women in the general population of Japan. Int Arch Occup Environ Health. 2000;73(3):163–70. pmid:10787131.
  51. 51. Ikeda M, Ezaki T, Tsukahara T, Moriguchi J, Furuki K, Fukui Y, et al. Reproducibility of urinary cadmium, alpha1-microglobulin, and beta2-microglobulin levels in health screening of the general population. Archives of environmental contamination and toxicology. 2005;48(1):135–40. pmid:15666443.
  52. 52. Akerstrom M, Barregard L, Lundh T, Sallsten G. Variability of urinary cadmium excretion in spot urine samples, first morning voids, and 24 h urine in a healthy non-smoking population: implications for study design. Journal of exposure science & environmental epidemiology. 2014;24(2):171–9. Epub 2013/09/12. doi: 10.1038/jes.2013.58 pmid:24022669.
  53. 53. Mason HJ, Williams NR, Morgan MG, Stevenson AJ, Armitage S. Influence of biological and analytical variation on urine measurements for monitoring exposure to cadmium. Occupational and environmental medicine. 1998;55(2):132–7. Epub 1998/06/06. pmid:9614399; PubMed Central PMCID: PMCPmc1757554.
  54. 54. Yamagami T, Suna T, Fukui Y, Ohashi F, Takada S, Sakurai H, et al. Biological variations in cadmium, alpha 1-microglobulin, beta 2-microglobulin and N-acetyl-beta-D-glucosaminidase in adult women in a non-polluted area. Int Arch Occup Environ Health. 2008;81(3):263–71. Epub 2007/10/19. doi: 10.1007/s00420-007-0206-z pmid:17943303.
  55. 55. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, et al. Reproducibility and Validity of a Semiquantitative Food Frequency Questionnaire. American journal of epidemiology. 1985;122(1):51–65. pmid:4014201
  56. 56. Jacques PF, Sulsky SI, Sadowski JA, Phillips JC, Rush D, Willett WC. Comparison of micronutrient intake measured by a dietary questionnaire and biochemical indicators of micronutrient status. The American journal of clinical nutrition. 1993;57(2):182–9. Epub 1993/02/01. pmid:8424386.
  57. 57. Keller A, Abbaspour KC, Schulin R. Assessment of uncertainty and risk in modeling regional heavy-metal accumulation in agricultural soils. Journal of environmental quality. 2002;31(1):175–87. Epub 2002/02/12. pmid:11837421.
  58. 58. Schaefer EJ, Augustin JL, Schaefer MM, Rasmussen H, Ordovas JM, Dallal GE, et al. Lack of efficacy of a food-frequency questionnaire in assessing dietary macronutrient intakes in subjects consuming diets of known composition. The American journal of clinical nutrition. 2000;71(3):746–51. pmid:10702168
  59. 59. Schatzkin A, Kipnis V, Carroll RJ, Midthune D, Subar AF, Bingham S, et al. A comparison of a food frequency questionnaire with a 24-hour recall for use in an epidemiological cohort study: results from the biomarker-based Observing Protein and Energy Nutrition (OPEN) study. International journal of epidemiology. 2003;32(6):1054–62. doi: 10.1093/ije/dyg264. pmid:14681273
  60. 60. Itoh H, Iwasaki M, Sawada N, Takachi R, Kasuga Y, Yokoyama S, et al. Dietary cadmium intake and breast cancer risk in Japanese women: a case-control study. International journal of hygiene and environmental health. 2014;217(1):70–7. Epub 2013/04/24. doi: 10.1016/j.ijheh.2013.03.010 pmid:23608001.
  61. 61. Julin B, Bergkvist C, Wolk A, Akesson A. Cadmium in diet and risk of cardiovascular disease in women. Epidemiology (Cambridge, Mass). 2013;24(6):880–5. Epub 2013/09/14. doi: 10.1097/EDE.0b013e3182a777c9 pmid:24030503.
  62. 62. Thomas LD, Elinder CG, Wolk A, Akesson A. Dietary cadmium exposure and chronic kidney disease: a population-based prospective cohort study of men and women. International journal of hygiene and environmental health. 2014;217(7):720–5. Epub 2014/04/03. doi: 10.1016/j.ijheh.2014.03.001 pmid:24690412.
  63. 63. Akerstrom M, Barregard L, Lundh T, Sallsten G. The relationship between cadmium in kidney and cadmium in urine and blood in an environmentally exposed population. Toxicology and applied pharmacology. 2013;268(3):286–93. Epub 2013/03/05. doi: 10.1016/j.taap.2013.02.009 pmid:23454399.
  64. 64. Fransson MN, Barregard L, Sallsten G, Akerstrom M, Johanson G. Physiologically-based toxicokinetic model for cadmium using Markov-chain Monte Carlo analysis of concentrations in blood, urine, and kidney cortex from living kidney donors. Toxicological sciences: an official journal of the Society of Toxicology. 2014;141(2):365–76. Epub 2014/07/13. doi: 10.1093/toxsci/kfu129 pmid:25015660.
  65. 65. Kjellstrom T, Nordberg GF. A kinetic model of cadmium metabolism in the human being. Environmental research. 1978;16(1–3):248–69. Epub 1978/07/01. pmid:679914.