Exploring potential dietary contributions including traditional ... - Nature

2 downloads 0 Views 178KB Size Report
Jan 24, 2007 - aNorth Queensland Health Equalities Promotion Unit, School of Medicine, ... Keywords: indigenous health, toxic metals, accumulation, kidney ...
Journal of Exposure Science and Environmental Epidemiology (2007) 17, 298–306 r 2007 Nature Publishing Group All rights reserved 1559-0631/07/$30.00

www.nature.com/jes

Exploring potential dietary contributions including traditional seafood and other determinants of urinary cadmium levels among indigenous women of a Torres Strait Island (Australia) MELISSA HASWELL-ELKINSa, VICTOR MCGRATHb, MICHAEL MOOREb, SOISUNGWAN SATARUGb, MARIA WALMBYa AND JACK NGb a

North Queensland Health Equalities Promotion Unit, School of Medicine, University of Queensland, Cairns, Queensland, Australia National Research Centre for Environmental Toxicology, Coopers Plains, Cairns, Queensland, Australia

b

Indigenous people of the Torres Strait Islands have been concerned about the safety of their traditional seafoods since the discovery of high cadmium levels in the liver and kidney of dugong and turtle in 1996. This study explored links between urinary cadmium levels and consumption frequency of these traditional foods and piloted a community-based methodology to identify potential determinants of cadmium exposure and accumulation. Consultations led to selection of one community for study from which 60 women aged 30 to 50 years participated in health and food frequency survey, urine collection and a routine health check. Urinary cadmium levels were determined by inductively coupled plasma-mass spectrometry; data were analysed using SPSS14. The geometric mean cadmium level in this group of women was 1.17 (arithmetic mean 1.86) mg/g creatinine with one-third exceeding 2.0 mg/g creatinine. Heavy smoking (Z300 pack years) was linked to higher cadmium in urine, as was increasing age and waist circumference. Analysis of ageadjusted residuals revealed significant associations (Po0.05) between cadmium level and higher consumption of turtle liver and kidney, locally gathered clams, peanuts, coconut, chocolate and potato chips. Dugong kidney consumption approached significance (P ¼ 0.06). Multiple regression revealed that 40% (adjusted r2) of variation in cadmium level was explained by the sum of these associated foods plus heavy smoking, age and waist circumference. No relationships between cadmium and pregnancy history were found. This paper presents a novel approach to explore contributions of foods and other factors to exposure to toxins at community level and the first direct evidence that frequent turtle (and possibly dugong) liver and kidney and wild clam consumption is linked to higher urinary cadmium levels among Torres Strait Islander women. Journal of Exposure Science and Environmental Epidemiology (2007) 17, 298–306. doi:10.1038/sj.jes.7500547; published online 24 January 2007

Keywords: indigenous health, toxic metals, accumulation, kidney disease, dietary sources, smoking.

Introduction National food standards and toxicological monitoring systems have been set up in most developed and many developing countries to ensure that populations are not exposed to foodborne toxins at levels that threaten health (e.g. FSANZ, 2003; JEFCA, 2003). For some toxins, a large safety margin is easily built into these standards as levels in foods are low relative to levels that threaten health. For toxic metals that are not excreted and accumulate over time, the margin between actual exposure and critical effect levels may be small (FSANZ, 2003; Jarup, 2003; Satarug and Moore, 2004), particularly for people at greater susceptibility (e.g. those with diabetes). Cadmium is a heavy metal present in food and cigarette smoke that accumulates and can cause damage to the kidney

1. Address all correspondence to: Dr. M. Haswell-Elkins, North Queensland Health Equalities Promotion Unit, 19 Aplin Street, PO Box 1103, Cairns, Queensland 4870, Australia. Tel.: þ 61-7-4046-8563. Fax: þ 61-7-4046-8555. E-mail: [email protected] Received 1 July 2006; accepted 7 November 2006; published online 24 January 2007

cortex. Although the critical effect level was initially estimated at 200 mg/g kidney cortex, recent studies have linked measurable effects on renal tubules at much lower levels (Buchet et al, 1990; Jarup et al., 1998; Jarup et al., 2000; Hellstrom et al., 2001; Satarug and Moore, 2004; Akesson et al., 2005). There is a relatively small gap between actual (24 to 42 mg per day globally) and the currently accepted ‘‘safe’’ intake of 70 mg per day (or 7 mg/kg body weight per week) (JEFCA, 2003). Recent estimates of average exposures in the general Australian population range from 13% to 68% of the provisional tolerable limit with the highest proportions in young children (FSANZ, 2003). In the Torres Strait Islands located between Papua New Guinea and the Australian mainland, dugong (a sea cow related to the manatee) and turtle represent important traditional foods of the Indigenous Islander people. In 1996 very high levels of cadmium were discovered in the liver and kidney of these long-lived animals (Gladstone, 1996). Subsequent dietary modelling indicated that consumption of these foods at levels realistic to their estimated availability in the community could result in cadmium consumption above the permitted tolerable weekly intake (Haynes and Kwan,

Dietary intake and cadmium levels

2001). A recent paper (Haswell-Elkins et al., in press) described increased cadmium levels in two Torres Strait Island communities associated with age, being female and being a smoker and suggestive links with having diabetes, a higher percentage body fat and living in a community with a high dugong and turtle catch. However, use of whole community comparisons to examine the contribution of dugong and turtle consumption is not ideal because consumption frequency may vary widely between individuals in the same community, interisland trading and shared feasting increases access of all Islanders to these foods and cadmium exposure can occur via other foods and smoking tobacco. Some data suggest that cadmium in seafood may not be highly bioavailable or absorbed because of competition with other essential metals (Bergland, et al., 1994; Vahter et al., 1996; Satarug et al., 2004; Johansen et al., in press). Higher levels of cadmium typically found among women appear to be due to their lower body iron stores compared to men, which could potentially be exacerbated by closely spaced pregnancies without iron supplementation (King, 2003). This project used a community-based approach to explore dietary, anthropometric and pregnancy factors, and age as potential determinants of urinary cadmium levels among Torres Strait Islander women.

Materials and methods Sampling Strategy and Rationale One Torres Strait Island community was selected from a total of 17 Outer Island communities for the following reasons:  Ample estimated catch of dugong and turtle so accessed would be fairly typical.  Chronic disease profile in relation to diabetes, renal disease and elevated albuminuria (micro and macro) similar to other Island communities.  Sufficient population to recruit 60 female participants in a narrow age range.  Community interest and desire to have a local study to learn more about health risks posed by cadmium in seafood and other sources. The age range of 30 to 50 years was selected because cadmium accumulates steadily with age until approximately age 50 when urinary cadmium levels begin to decline (HaswellElkins et al., in press; WHO, 1992). The study focused on women because of their higher risk of exposure, absorption and health effects from cadmium (WHO, 1992; Satarug et al., 2000; Choudhury et al., 2001; Satarug et al., 2004).

Recruiting Participants and Implementing the Questionnaire Extensive consultations were conducted with the Health Council and District Health Service, the Island Coordinating Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Haswell-Elkins et al.

Council and the specific Island Council and health team to ensure understanding of issues among local, regional and community leaders. These consultations revealed that people wanted more directly relevant information on the potential effects of cadmium on their health. The study was welcomed by all those consulted and was approved by Ethics Committees of the University of Queensland and Queensland Health. The field research team consisted of an epidemiologist with toxicology and Indigenous health experience, a highly respected Torres Strait Islander researcher with a long involvement in environmental work and an Indigenous university student. A list of all resident women (total 75) within the age range 30 to 50 years from the health information system was provided to the field team who worked closely with the Primary Health Centre staff. Each was visited individually, and the study was explained in detail to ensure full understanding of the purpose and process. Informed written consent was provided by 60 women. Eleven women were away or could not be located and four declined participation. A questionnaire covering areas of relevance to cadmium health effects was developed and piloted before arrival. The survey was administered orally and took approximately 45 min to complete. Questions addressed general health, chronic disease, history of pregnancy, smoking, iron tablets, selected food frequencies, knowledge and interest about kidneys and kidney care. Participants were asked about their present and past smoking experience. Current smokers were asked how many years they had been smoking, how many cigarettes they smoked each day and whether there had been significant changes in smoking over time. Non-smokers were asked if they had ever been a smoker; if so they were asked about their previous smoking experience. Pack years were calculated (number of smoking years times usual number of cigarettes smoked per day). Those scoring 300 or more were considered heavy smokers, those who smoked less were coded as light to moderate and those who never smoked were coded as never-smokers. Dietary information was collected with emphasis on foods that have been identified or suspected as potentially higher in cadmium, some control foods not expected to be significant sources of cadmium and some with high iron content that could possibly be protective against cadmium absorption. The potential cadmium-containing foods assessed were beef kidney, seafoods, particularly the kidney and liver of the dugong and turtle, turtle intestine, clams, oysters and other locally popular shellfish, peanuts, potato chips (packaged, potato crisps), chocolate and coconut and coconut milk. Control foods included dugong meat, dugong intestine, a range of green vegetables and fruit, potatoes, sweet potatoes and red meat. Table 1 provides estimates of cadmium levels in these foods determined from other studies (Marro, 1994; 299

Dietary intake and cadmium levels

Haswell-Elkins et al.

Table 1. Average (geometric mean) levels of cadmium among participants grouped according to frequency of consumption of particular food items. Food item (estimated cadmium concentration (mg/kg) from other studies (references))

Geometric mean (mg Cd/g creat) Low# (n)

Age-adjusted residuals P-value

High (n)

Infrequent foods # low ¼ never to few times per year; high ¼ monthly or more Dugong liver 3.66 (1,4,5) 1.07 (50) Dugong kidney 8.52 (1,4,5) 1.00 (48) Turtle liver 12.82 (1,4,5) 0.83 (35) Turtle kidney 26.00 (4,5) 0.92 (45) Akul shellsa 0.44 (1,4) 1.11 (43) Pippi shellsa 0.44 (1,4) 1.09 (41) Dugong meat, fat 0.01 (1,4) 0.96 (29) Dugong intestine 0.09 (4) 1.08 (40) Wild clams Relevant estimates N/A 0.97 (51) Beef kidney N/A (MPC, 2.5 mg/kg) (2) 1.20 (43) Medium frequency foods: # low ¼ never to monthly; high ¼ weekly Peanuts 0.013–0.031(5) (MPC 0.1 mg/kg) Chocolate 0.017 (2) (MPC chocolate/cocoa products 0.5 mg/kg) Potato chips 0.135 (2) Oysters 0.419 (3) (MPC 2.0 mg/kg) Turtle meat 0.04 (1) Turtle intestine 3.66 (2) Crayfish heads 0.11/14.5b(1)

Not age-adjusted P-value

1.79 (10)

0.20

0.22

2.18 (12)

0.03*

0.06

1.88 (25)

0.005**

0.01**

2.39 (15)

0.004**

0.05*

1.33 (17)

0.59

0.91

1.37 (19)

0.46

0.44

1.42 (31)

0.20

0.19

1.38 (20)

0.44

0.53

3.33 (9)

0.002***

0.02*

1.30 (17)

0.68

0.72

0.74 (29)

1.71 (30)

0.003**

0.002**

1.01 (41)

1.44 (19)

0.26

0.05*

0.97 (28)

1.31 (31)

0.66

0.05*

1.13 (47)

1.31 (13)

0.69

0.96

1.16 (46)

1.22 (14)

0.90

0.56

1.12 (50)

1.44 (10)

0.53

0.42

1.20 (29)

1.29 (30)

0.79

0.76

0.74 (18)

0.04*

0.16

0.96 (31)

0.22

0.30

1.19 (32)

0.75

0.65

1.12 (25)

0.93

0.86

1.11 (25)

0.89

0.74

2.01 (17)

0.01*

0.01*

or more

High frequency foods: #low ¼ never to once weekly; high ¼ many times weekly Red meat Undetectable (2,3) (MPC 0.05 mg/kg) 1.42 (42) Potato 0.029 (2) (MPC 0.1 mg/kg) 1.37 (29) All fruit Most o0.005 (2,3) 1.08 (28) All vegetables Most o0.01 (2,3) 1.15 (35) Sweet potato N/A; (MPC 0.01) (2) 1.15 (35) Coconut N/A 0.91 (43)

Items that are starred (*) indicate a statistically significant difference in average cadmium level between relatively low and high frequency consumers. P-values with and without control for age are provided. N/A, not available. References for the cadmium concentration estimations: 1 ¼ Haynes and Kwan (2001); 2 ¼ FSANZ (2003); 3 ¼ Marro (1996); 4 ¼ Gladstone (1996); 5 ¼ Gordon et al. (1998); 6 ¼ Tinggi (1998); All MPCs reported in table from Australia New Zealand Food Standards Code: Standard 1.4.1. a No specific levels available, but mangrove cockles may be considered proxy, 0.44 mg/kg (1). b Without and with hepatopancreas, respectively.

300

Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Dietary intake and cadmium levels

Gladstone, 1996; Gordon et al., 1998; Tinggi, 1998; Haynes and Kwan, 2001; FSANZ, 2003). Substantial concentrations of cadmium were found in kidney tissue from clams (species Triakcna) gathered and eaten locally (Dight and Gladstone, 1993), but whole animal estimates are not available. Pregnancy histories captured information about the number and intervals between pregnancies, total duration of child-bearing years and whether iron tablets were taken during the time. Total years of pregnancy was estimated from this information and used in the analysis. Initial frequency analysis of the monitored foods showed that some are typically eaten on a daily to weekly basis (red meat, white and sweet potatoes, green vegetables, fruits, coconut and crayfish) whereas others are much less frequently consumed, monthly or a few times per year. The latter includes the organ meat of dugong and turtle, clams and chocolate containing foods. Reflecting this, food frequency categories were combined into larger groups in order to enable statistical analysis (see Table 1 for grouping criteria). Infrequent foods considered consumption a few times a month or more to be relatively high frequency, medium frequency foods were considered high on a weekly basis or more, whereas high frequency foods were grouped according to weekly or less versus several times a week. Several elders were also interviewed to share understanding of cultural significance linked to eating dugong and turtle organs, to explore their views about heavy metals in these foods and to ensure they were aware of reasons, and later the results, of the study. For background health data, Primary Health Centre staff conducted a routine recall of all women in the age range to complete their annual health check, including measurement of height, weight, waist circumference, blood pressure, random blood sugar and albumin creatinine ratio. Urine was collected in large urinals. The time between the previous and the collected urines and total volume were recorded. Specimens were transported on ice and samples were preserved in four tubes including one precoated with a drop of concentrated nitric acid (HNO3) then frozen as soon as possible. Frozen samples were transported to the National Research Centre for Environmental Toxicology for analysis. Cadmium levels were measured in urines preserved with HNO3 by ICP-MS (inductively coupled plasma-mass spectrometry) with appropriate QA/QC samples. Briefly, 0.5 ml of urine and 0.1 ml of 500 ng/ml internal standard mix was added to 4.25 ml 2% HNO3, centrifuged and analysed using ICP-MS (Agilent 7500CS). Four urine samples were spiked with extra 10 mg cadmium/l in the assay solutions. The average recovery from the spiked samples was 9371% (93, std 1%, n ¼ 4). Certified reference urine material (Lyphocheck-1-low level and Lyphocheck-2-high level, Bio-Rad, CA, USA) was also included. The average cadmium levels were found to be 5.870.1 mg/l (n ¼ 4) and 11.470.1 mg/l Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Haswell-Elkins et al.

(n ¼ 4) with the expected certified averages of 5.9 mg/l (acceptable range 4.7 to 7.1) and 11.9 mg/l (acceptable range 9.5 to 14.3) for Lyphocheck-1 and Lyphocheck-2 respectively. When the results were compared to the average certified values, the recoveries were 98% and 95% for the two reference urine materials, respectively. Creatinine levels were determined using an end point colorimetric assay kit. Absorbance was measured using a spectrophotometer at 550 nm. A standard was run every 10 samples and every tenth was carried out in duplicate for quality control. The assay was linear up to 150 mg/dl.

Statistical Analysis of the Data All data from the questionnaires, the background health checks, cadmium determinations and body measurements were coded where needed and entered onto SPSS for analysis. Data were tested for normal distribution using the Kolmogorov–Smirnov test and log-transformed to achieve normality were necessary. Statistical tests used to identify relationships between these variables included w2 testing of cross-tabulated data, Pearson’s correlation, analysis of variance and linear regression. In order to control for the linear relationship between age and cadmium level, unstandardised residuals created by regression of cadmium level against age were tested with each food item. The final analysis explored whether the additive nature of cadmium intake from the various associated foods identified individually was demonstrable. This was carried out by computing the sum of associated foods and/or risks for each individual out of a possible nine, namely relatively high consumption of turtle liver, turtle kidney, dugong kidney, clams, coconut, peanuts, chocolate and potato chips plus heavy smoking. A second group of sums was calculated for associated land foods (peanuts, chocolate, coconut, potato chips) and seafoods (turtle liver and kidney, dugong kidney, clams) and heavy smoking. These sums, together with age and waist circumference, were tested in linear regression models to assess relationships with variation in urinary cadmium level.

Results Levels of Cadmium in the Sample The average, geometric mean and median cadmium levels for the group of 60 women from the Island was 1.86 (SE mean 0.22), 1.17 and 1.47 mg of cadmium per gram of creatinine, respectively. Levels ranged from 0.03 to 7.82 mg/g creatinine. The values were distributed with roughly one-third (38.3%, 24 women) between 0 and 0.99, 30% (17 women) between 1 and 1.99 and 31.7% (19 women) at 2.0 mg/g creatinine or above. As expected, age was strong determinant of cadmium level (Figure 1, ANOVA F ¼ 8.3, df 3,56, Po0.001) and is considered in further analysis. 301

Dietary intake and cadmium levels

4.5

5

4 3.5 3 2.5 2 1.5 1 0.5 0

30-34

35-39

40-44

45-50

Age Group (Years)

Urinary Cadmium Geometric Mean (µg/g creatinine)

Urinary Cadmium Geometric Mean (µg/g creatinine)

Haswell-Elkins et al.

4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

Figure 1. Geometric mean cadmium levels and 95% confidence intervals in groups of participating women stratified by age in years. Sample sizes in each group are 14 (30 to 34), 15 (35 to 39), 17 (40 to 44) and 14 (45 to 50).

Liver

Kidney

302

Clams

Yearly

Monthly to Weekly

Figure 2. Geometric mean urinary cadmium levels (and 95% confidence intervals) in groups stratified by reported frequency of consumption of dugong and turtle liver and kidney and clams. Sample sizes: 40, 10, 10 for dugong liver; 40, 8, 12 for dugong kidney; 22, 13 and 25 for turtle liver; 40, 5, 15 for turtle kidney; 30, 21 and 9 for clams.

5 4.5 Urinary Cadmium Geometric Mean (µg/g creatinine)

Likely Sources of Variation in Cadmium Levels F Foods Table 1 shows the average urinary cadmium levels among women reporting relatively low versus high consumption frequency for each food assessed. Foods associated with statistically significant differences in urinary cadmium levels according to consumption frequencies are shown in bold. Women who reported a relatively high frequency (monthly or more) of consumption of dugong kidney (but not dugong liver), turtle liver, turtle kidney and clams (Figure 2), and consumed peanuts weekly or more and coconut or coconut milk several times a week or more (Figure 3) had significantly higher urinary cadmium levels than those who consumed these foods less frequently (Table 1). In contrast, those who consumed potatoes and red meat many times a week had lower cadmium levels; the difference reached statistical significance for red meat (P ¼ 0.04). Average values of unstandardised residuals created by linear regression of age against cadmium level were compared between relatively low and high consumers of each food item to control for the possibility that eating patterns of some items vary by age. Table 1 also shows the probability value for statistical significance in cadmium levels associated with low and high consumption of food items controlling for age. This tended to reduce, but in most cases not eliminate, the significance of associations with traditional foods, for example, for dugong kidney, turtle liver and kidney and clams. Adjustment for age did not affect the significance associated with consumption of some land foods (peanuts

Kidney Turtle

Traditional Food Never

In this study, two ways of expressing cadmium were used. One was by adjusting cadmium per gram of creatinine to correct for urine dilution and the second was by estimating the whole amount of urine excreted in a known time period. These two measures correlated closely (Pearson’s r ¼ 0.75, Po0.001) overall and among women with diabetes (r ¼ 0.83, n ¼ 14) and those without (r ¼ 0.72, n ¼ 46). As cadmium adjusted for creatinine is more commonly used, these were used in the analysis.

Liver

Dugong

4 3.5 3 2.5 2 1.5 1 0.5 0 Peanuts Never to Yearly

Coconut Monthly

Weekly

Many Times Weekly

Figure 3. Geometric mean cadmium levels (and 95% confidence intervals) in groups stratified by frequency of consumption of coconut and coconut milk and peanuts. Sample sizes: coconut (9, 16, 16, 17); peanuts (15, 14, 23, 7).

and coconut), but revealed relationships with chocolate and potato chips that were not previously visible. The negative association with red meat consumption mentioned above was attenuated by inclusion of age. Table 2 examines the frequency distributions of cadmium levels (into o1, 1 to 2 and 42 mg/g creatinine) among women who reported high consumption of none, one or two to four of the seafoods identified to be linked to higher cadmium level. Seventeen per cent of the women with no higher consumption of associated seafoods excreted cadmium over 2 mg/g creatinine in contrast to 67% of those who were frequent consumers of two to four of the seafoods. Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Dietary intake and cadmium levels

Haswell-Elkins et al.

Table 2. Seafood risks (sum of dugong kidney, turtle liver and kidney, clam). Number of risks

% low cadmium (0–o1.0 mg/g creat)

0 1 2 to 4

14 7 2

% medium cadmium (1–o2 mg/g creat)

48.3 43.8 13.3

10 5 3

34.5 31.3 20.0

% high cadmium (42 mg/g creat) 5 4 10

17.2 25.0 66.7

Highly significant w2 linear by linear association, P ¼ 0.001.

Urinary Cadmium Geometric Mean (µg/g creatinine)

Urinary Cadmium Geometric Mean (µg/g creatinine)

4 3.5 3 2.5 2 1.5 1 0.5

4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 0

0 Never

Light to Moderate

Heavy

Smoking History

Figure 4. Geometric mean cadmium levels and 95% confidence intervals among the 60 women stratified by groups who never smoked (n ¼ 22), for those who smoked o300 pack years (n ¼ 25) and those with 4300 pack years of exposure (n ¼ 13).

Other Factors Among the 60 women, 22 (37%) had never ‘‘been a smoker’’ and 52% reported currently smoking. Although mean cadmium levels varied little between never smokers and light to moderate smokers, higher cadmium levels among heavy smokers (Z300 pack years; Figure 4) approached significance (F ¼ 3.8, Po0.06). The addition of age in the analysis, however, showed a loss of significance associated with heavy smoking (F ¼ 1.7, df 3, 52, P ¼ 0.20). The average waist circumference of the participating women was 109.9 cm (standard deviation ¼ 14.0). Waist circumference, but not body mass index or weight, was significantly associated with cadmium levels (Pearson’s r ¼ 0.30, df 60, P ¼ 0.02) and control for age did not affect this relationship (partial correlation 0.29, P ¼ 0.03). None of the pregnancy-related information, that is number, average interval between pregnancy, total duration, iron tablets, smoking in pregnancy or gestational diabetes experience showed significant links with cadmium level in any statistical testing (data not shown). Additive Models An incremental increase in mean cadmium levels among women categorised by numbers of postulated significant exposure factors is shown in Figure 5. The linearity of this relationship is highly significant (log-transformed cadmium Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

1 2 3 4 Number of Cadmium Associated Factors

5-8

Figure 5. Geometric mean cadmium levels and 95% confidence intervals in groups stratified by the number of factors associated with cadmium level (out of a total of 9) identified in this study experienced by the individual. Sample sizes for each risk category: 0 ¼ 10; 1 ¼ 6; 2 ¼ 12; 3 ¼ 10; 4 ¼ 8; and 5 to 8 ¼ 12. Two samples had missing information and were excluded. Cadmium-associated factors included relatively high consumption of turtle liver, turtle kidney, dugong kidney, clams, coconuts, peanuts, potato chips, chocolate and heavy smoking.

level, Linearity F ¼ 27.3, df 1, 57; Po0.001) and remained strong using unstandardised residuals of cadmium adjusted for age (F ¼ 30.2, df 1, 57; Po0.001). Linear regression models (Table 3) confirmed a significant contribution of both seafoods and land foods, as well as age and waist circumference in the explanation of variation in cadmium levels among the women. Overall these model accounted for 40% of the total variation.

Discussion This small study revealed strong relationships between urinary cadmium levels and dietary intake of certain foods not only those predicted to cause markedly high exposure (such as dugong and turtle liver and kidney) but also some previously unknown possible sources (e.g. wild clams, coconut products) and some everyday foods (peanuts, potato chips, chocolate). Thus, this paper details a useful method to explore determinants of urinary cadmium using an empirical approach that has provided potentially important findings to guide further research. Among the 60 participating women between the age of 30 and 50, 32% excreted over 2 mg of cadmium per gram 303

Dietary intake and cadmium levels

Haswell-Elkins et al.

Table 3. Details of multiple regression analysis using forward stepwise entry of variables to examine variation in urinary cadmium levels (log transformed mg per gram creatinine). Variable

Unstandardised B (SE)

Standardised beta

T (P-value)

R (adjusted r2) with addition

Model 1 Using a combined score for associated seafood, land foods and smoking Constant 3.40 (1.82) Sum of nine associated factors 0.11 (0.02) 0.50 Age 0.02 (0.01) 0.30 Waist circumference 2.06 (0.89) 0.24

1.87 4.85 2.93 2.31

(0.07) (0.001) (0.005) (0.025)

0.59 (0.32) 0.64 (0.09) 0.68 (0.05)

Model 2. Separate scores for associated sea and land foods and heavy smoking Constant 3.87 (1.79) Seafoods (4) 0.090 (0.04) 0.239 Land foods (4) 0.212 (0.06) 0.399 Age 0.025 (0.01) 0.312 Waist circumference 2.26 (0.87) 0.260

2.16 2.17 3.79 2.91 2.60

(0.04) (0.04) (o0.001) (0.005) (0.01)

0.47 0.58 0.65 0.70

Variables not entered into Model 2. Heavy smoking Beef consumption

0.06 0.16

creatinine, and 11.7% excreted levels above 4.0. These levels are similar to that found among women of this age range in a previous study in 1996 of urine samples collected from two other Torres Strait Islands (Haswell-Elkins et al., in press). As expected, age was an important determinant of cadmium levels. As the metal cannot be effectively excreted from the body, it slowly collects in the cortex of the kidney and steadily increases with age. A very small percentage escapes continuously in the urine and levels in urine reflect the levels in the kidney cortex (Nordberg and Nordberg, 1988; WHO, 1992; Trevisan et al., 1994; Choudhury et al., 2001). This study presents the first direct evidence linking consumption of turtle liver and kidney, wild gathered clams and (suggested but not statistically significant) dugong kidney with cadmium levels among Torres Strait Islanders. The data suggest that, at least in the case of turtle organs, the cadmium is bioavailable and absorbed from these foods by the body contributing to body burden and urinary levels. These data support the modelling by Haynes and Kwan (2001), which indicated that quantities of turtle and, to a lesser extent, dugong tissue available for consumption on some Torres Strait Islands is sufficient to cause residents to exceed the permitted tolerable weekly intake of cadmium. Surprisingly, weak associations were observed between cadmium and smoking experience, despite the collection and examination of detailed data. The reason for this remains unclear as smoking was both prevalent and relatively heavy among the participants. Differences in cadmium levels associated with smoking may be more difficult to detect when dietary contributions and variability are relatively high. This study found positive associations between cadmium level and waist circumference after controlling for many variables; similar to our previous observations with body fat (Haswell-Elkins et al., in press). Obesity is very common in 304

(0.22) (0.12) (0.09) (0.07)

0.52 (0.61) 1.5 (0.14)

the Torres Strait suggesting that food energy intake frequently exceeds body requirements (Leonard et al., 2002). As kidney size does not increase with excess intake, the renal concentration of cadmium may increase because of greater exposure to larger quantities of food. As mentioned above, the approach used in this project to identify potential risk foods for cadmium through population level studies is different from the most commonly used methods. Food safety assessments usually involve working from the level of the toxin in the food times the frequency of consumption in the population times the number of years the toxin accumulates (FSANZ, 2003; JEFCA, 2003; Judd, et al., 2004) with estimations for high consumers. Although this assessment method has major benefits, it may not adequately capture variability in consumption experienced by groups with different eating patterns and extremely limited dietary variety (Kim, et al., 1998; Kuhnlein and Chan, 2000; Lee et al., 2002; Judd et al., 2004). Populationbased studies of urinary cadmium can directly measure (rather than infer) the distribution of cadmium levels in populations. This study demonstrates their potential in helping to identify individual and groups of foods associated with cadmium level. Another limitation of national dietary and market basket surveys is that foods that are not in the commercial system go unnoticed and people may be unaware of unhealthy exposures (Kim et al., 1998; Kuhnlein and Chan, 2000; Johansen et al., 2004). Traditional foods gathered only, or more frequently, by specific cultural groups may be known to contain high levels of toxin, but they fall outside consideration of food safety regulation. Although oysters in this study are mainly purchased in the store, clams (which were linked with higher cadmium levels among consumers) and coconuts are frequently gathered locally. There are many similar Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Dietary intake and cadmium levels

situations where there is concern over levels of cadmium and other toxins in Indigenous traditional foods, for example oysters, clams, crabs, sea mammal organs, plants and fungi, moose and caribou organs by First Nations and Inuit peoples in Canada and Greenland (Kuhnlein and Chan, 2000; Johansen et al., 2004). There are many limitations of this study, for example its small size, its reliance on estimations based on general recall of usual consumption and it did not attempt to correct for dietary changes over time. However on remote islands, the limited variety of foods available minimises variation in diet across the community (Lee et al., 2002). Perhaps most importantly, this study had little power to distinguish with precision which foods independently contributed to cadmium levels and which were indicators of a dietary pattern. Although consumption frequencies with most foods were linearly associated with cadmium levels, this was not the case for peanuts (Figure 3). It seems unlikely that peanuts (and possibly coconut) are contributing significant amounts of cadmium to the diet; since FSANZ (2006) estimated that only 4% of the total amount of cadmium in the general population is linked to peanut consumption. Thus, the links between peanut (and possibly coconut) consumption and cadmium level might reflect a particular dietary pattern (more traditional) that is richer in all types of seafood than people who more frequently eat, for example, potatoes and red meat. There appears to be little available data on cadmium levels in coconut and coconut products. Further research is needed to clarify these issues. In conclusion, the levels of cadmium among the women in this study were moderately high with at least one-third excreting over 2 mg cadmium per gram creatinine and at possible risk of renal damage, especially in combination with diabetes (Buchet et al, 1990; Jarup et al. 1998; 2000; Hellstrom et al., 2001; Satarug and Moore, 2004; Akesson et al., 2005). Although there are limitations in the data, the strength of the associations observed and their ability to address concerns raised by the Elders and women involved, especially for informing their children about turtle and dugong liver and kidney and wild clam consumption, add to the growing body of literature enabling toxicological research to help meet people’s information needs about these issues (Kim et al., 1998; Kuhnlein and Chan, 2000; Johansen et al., 2004; Judd et al, 2004).

Acknowledgements This research was funded in part by a grant from Australian Institute for Aboriginal and Torres Strait Islander Studies, Canberra. We thank the Torres Strait and Northern Peninsula Area Health Council and Health District, and the Community Council and Health Centre of the study island. We particularly thank Phillip Mills (District Manager) and Associate Professor Peter O’Rourke for support. Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)

Haswell-Elkins et al.

References Akesson A., Lundh T., Vahter M., Bjellerup P., Lidfeldt J., and Nerbrand C., et al. Tubular and glomerular kidney effects in Swedish women with low environmental cadmium exposure. Environ Health Perspect 2005: 113: 1627–1631. Bergland M., Akesson A., Nermell B., and Vahter M. Intestinal absorption of dietary cadmium in women depends on body iron stores and fiber intake. Environ Health Perspect 1994: 102: 1058–1066. Buchet J.P., Lauwreys R., Roels H., Bernard A., Bruaux P., and Claeys F., et al. Renal effects of cadmium body burden of the general population. Lancet 1990: 336: 699–702. Choudhury H., Harvey T., Thayer W.C., Lockwood T.F., Stiteler W.M., and Goodrum P.E., et al. Urinary cadmium elimination as a biomarker of exposure for evaluating a cadmium dietary exposure F biokinetics model. J Toxicol Environ Health A 2001: 63: 321–350. Dight I., and Gladstone W. Torres Strait Baseline Study: Pilot Study Final Report June 1993. Trace metal concentration in sediment and selected marine biota as indicator organisms and food items in the diet of Torres Strait Islanders and Coastal Papuans, Great Barrier Marine Park Authority, 1993. Food Standards Australia New Zealand (FSANZ). The 20th Australian Total Diet Survey: A Total Diet Survey of Pesticide Residues and Contaminants, FSANZ, Canberra, 2003. Food Standards Australia New Zealand (FSANZ). Cadmium in Peanuts 2006 Website page: http://www/foodstandards.gov.au/mediareleasespublications/ factsheets1999/cadmiuminpeanuts.cfmdownloaded on January 5 2006. Gladstone W. Trace metals in sediments, indicator organisms and traditional seafoods of the Torres Strait. Final Report of the Torres Strait Baseline Study Great Barrier Reef Marine Park Authority, Townsville, Qld, Australia, 1996. Gordon A.N., Pople A.R., and Ng J. Trace metal concentrations in livers and kidneys of sea turtles from south-eastern Queensland, Australia. Mar Freshwater Res 1998: 49: 409–414. Haswell-Elkins M., Imray P., Satarug S., Moore M.R., and O’Dea K. Urinary excretion of cadmium among Torres Strait Islanders (Australia) at risk of elevated dietary exposure through traditional foods. J Expos Sci Environ Epidemiol, (in press). Haynes D., and Kwan D. Trace Metal Concentrations in the Torres Strait Environment and Traditional Seafood Species 1997–2000. Torres Strait Regional Authority, Thursday Island, Queensland, Australia, 2001 pp. 91. Hellstrom L., Elinder C.G., Dahlberg B., Lundberg M., Jarup L., Persson B., and Axelson O. Cadmium and end-stage renal disease. Am J Kidney Dis 2001: 38: 1001–1008. Jarup L. Hazards of heavy metal contamination. Br Med Bull 2003: 68: 167–182. Jarup L., Berglund M., Elinder C., Nordberg G., and Vahter M. Health effects of cadmium exposure F a review of literature and a risk estimate. Scand J Work Environ Health 1998: 24(suppl): 1–52. Jarup L., Hellstrom L., Alfven T., Carlsson M.D., Grubb A., and Persson B., et al. Low level exposure to cadmium and early kidney damage: the OSCAR study. Occup Environ Med 2000: 57: 668–672. Joint FAO/WHO Expert Committee on Food Additives (JEFCA) Sixty-first meeting 2003 Rome, 10–19 June 2003. Johansen P., Muir D., Asmund G., and Riget F. Human exposure to contaminants in the traditional Greensland diet. Sci Total Environ 2004: 331: 189–206. Johansen P., Mulvad G., Pedersen H.S., Hansen J.C., and Riget F. Accumulation of cadmium in livers and kidneys in Greenlanders. Sci Total Environ 2006: 372: 58–63. Judd N.L., Griffith W.C., and Faustman E.M. Consideration of cultural and lifestyle factors in defining susceptible populations for environmental disease. Toxicology 2004: 198: 121–133. Kim C., Chan H.M., and Receveur O. Risk assessment of cadmium exposure in Fort Resolution, Northwest Territories, Canada. Food Additives Contam 1998, 307–317. King J.C. The risk of maternal nutritional depletion and poor outcomes increases in early or closely spaced pregnancies. J Nutr 2003: 133(5 Suppl 2): 1732S–1736S. Kuhnlein H.V., and Chan H.M. Environment and contaminants in traditional food systems of northern Indigenous peoples. Annu Rev Nutr 2000: 20: 595–626. Lee A.J., Darcy A.M., Leonard D., Groos A.D., Stubbs C.O., and Lowson S.K., et al. Food availability, cost disparity and improvement in relation to accessibility and remoteness in Queensland. Aust NZJ Public Health 2002: 26: 266–272.

305

Haswell-Elkins et al.

Leonard D., McDermott R., and Odea K. Obesity, diabetes and associated cardiovascular risk factors among Torres Strait Islander people. Aust NZ J Public Health 2002: 26: 144–149. Marro N. 1994 Australian Market Basket Survey, Australian Government Publishing Service, Canberra, 1996. Nordberg G.R., and Nordberg M. Biological monitoring of cadmium. In: Clarkson TW, Friberg L, Nordberg G and Sager PR (Eds.) Biological Monitoring of Toxic Metals. New York, Plenum Press, 1988 pp. 151–168. Satarug S., Haswell-Elkins M.R., and Moore M. Safe levels of cadmium intake to prevent renal toxicity in human subjects. Br J Nutr 2000: 84: 791–802. Satarug S., and Moore M.R. Adverse health effects of chronic exposure to lowlevel cadmium in foodstuffs and cigarette smoke. Environ Health Perspect 2004: 112: 1099–1103.

306

Dietary intake and cadmium levels

Satarug S., Ujjin P., Vanavanitkun Y., Baker J.R., and Moore M.R. Influence of body iron store status and cigarette smoking on cadmium body burden of healthy Thai men and women. Toxicol Lett 2004: 148: 177–185. Tinggi U. Cadmium levels in peanut products. Food Additives Contam 1998: 15: 789–792. Trevisan A., Nicoletto G., Maso S., Grandesso G., Odynets A., and Secondin L. Biological monitoring of cadmium exposure: reliability of spot urine samples. Int Arch Occup Environ Health 1994: 65: 373–375. Vahter M., Berglund M., Nermell B., and Akesson A. Bioavailability of cadmium from shellfish and mixed diet in women. Toxicol Appl Pharmacol 1996: 136: 332–341. World Health Organisation Environmental Health Criteria No 134, Cadmium WHO, Geneva (ISBN 9241517349), 1992, 280 pp.

Journal of Exposure Science and Environmental Epidemiology (2007) 17(3)