Urinary Cadmium, Impaired Fasting Glucose, and Diabetes in the NHANES III

The National Center for Health Statistics of the Centers for Disease Control and Prevention conducted the NHANES III in 1988–1994 in a nationwide probability sample of ∼39,000 noninstitutionalized U.S. civilians aged 2 months and older (7,8). Information on demographic characteristics, ethnicity, and medical history of diabetes was obtained in a household interview. Information on history of diabetes included questions about prior diagnoses of diabetes by a physician and current use of insulin and oral hypoglycemic agents. In addition, women were asked whether the diagnosis had been made during pregnancy and whether they had been diagnosed with diabetes at some time other than pregnancy. Blood and urine specimens were obtained during physical examination. Urine was collected as a “spot” or untimed sample. Studies of timed urine specimens indicate that urinary cadmium levels do not show significant diurnal variation (9) and that urinary cadmium in spot specimens is well correlated with urinary cadmium detected in 12-h and 24-h specimens (10).

Plasma glucose, measured using a modified hexokinase enzymatic method (8), was used to classify participants under the headings “normal,” “impaired fasting glucose,” and “diabetes.” Subjects with serum cotinine ≤10 ng/ml and who reported smoking fewer than 100 cigarettes in their lifetimes were classified as nonsmokers; otherwise, subjects were classified as ever-smokers. Serum cotinine was measured using high-performance liquid chromatography coupled with an atmospheric pressure chemical ionization tandem mass spectrometer (11). Urine cadmium was measured by Zeeman effect graphite furnace atomic absorption (12), and urine microalbumin was measured using a solid-phase fluorescence immunoassay (13). To adjust for variation in the diluteness of urine, urinary cadmium levels were expressed as urine cadmium/urine creatinine (μg/g). Details regarding the laboratory procedures for these tests are published elsewhere (8).

Subjects were categorized as having impaired fasting glucose (IFG) and diabetes based on fasting (8- to 24-h) plasma glucose (FPG) levels in accordance with the criteria of the American Diabetes Association (14). IFG was defined as 110 ≤ FPG < 126 mg/dl; diabetes was defined as FPG ≥126 mg/dl and/or current use of insulin or oral hypoglycemic agents.

We restricted the analysis to adults 40 years of age and older to exclude cases of type 1 diabetes, which is etiologically distinct from type 2 diabetes. After excluding those with missing values, we analyzed data for the 8,722 subjects for whom there were complete records. We quantified the associations between urinary cadmium and IFG, diabetes, and the combined outcomes (IFG + diabetes) by estimating odds ratios (ORs) and calculating 95% CIs by logistic regression. The cadmium concentration in urine was categorized in three levels: 0–0.99 (reference), 1–1.99 (exposure level I), and ≥2 μg/g creatinine (exposure level II). Exposure levels I and II were modeled as two indicator variables. The ORs were adjusted for age, race, sex, and BMI. The relationship between urinary cadmium (μg/g creatinine) and smoking (ever-smokers versus nonsmokers) was examined using the Wilcoxon-Mann-Whitney test.

The prevalence of IFG, diabetes, and the combined outcomes (IFG + diabetes) were positively associated with urinary cadmium in a dose-dependent manner (Table 1). The odds of the combined outcomes increased by ∼30% at each level of urinary cadmium. The magnitudes of these associations were slightly higher for IFG than for diabetes. Similar effects were observed when urinary cadmium was modeled as a continuous variable; the ORs associated with an increase of 1 μg/g creatinine were 1.19 (95% CI 1.08–1.31) and 1.11 (95% CI 1.03–1.20) for IFG and diabetes, respectively.

Because cadmium is known to exacerbate renal damage in diabetes (15), renal damage could cause cadmium to leak into urine, leading to a (noncausal) association between cadmium and diabetes. We therefore restricted the analysis to persons without laboratory evidence of renal damage, using a standard cut-point of urine albumin, ≤30 μg/ml, as described by Paschal et al. (6). This restriction did not appreciably affect our findings (Table 2). Thus, the higher levels of cadmium in the urine of persons with IFG and diabetes do not appear to be the result of renal impairment.

Apart from occupational exposure to cadmium (which should be rare in this cohort), the major source of cadmium exposure among nonsmokers is the diet (16). Among smokers, the major source of cadmium exposure is cigarettes. As has been observed repeatedly (6,16), urinary cadmium levels were significantly higher among ever-smokers: the mean levels of urinary cadmium were 0.92 and 0.63 μg/g creatinine (ever-smokers versus nonsmokers; P < 0.0001). We examined the prevalence of the combined outcomes (IFG + diabetes) in ever-smokers and nonsmokers. Despite small variations in effect size at the different levels of exposure, the associations were apparent in both groups (Table 3). Thus, regardless of the source of exposure, diet or cigarettes, higher levels of cadmium in urine were associated with increases in the odds of IFG and diabetes. Finally, we tested the association between smoking and the combined outcomes: the OR adjusted for age, race, sex, and BMI was 1.12 (95% CI 1.00–1.25).

To our knowledge, this is the first report of an association between cadmium exposure and both IFG and type 2 diabetes. Because this association is observed among persons without evidence of renal damage and among persons with IFG (i.e., prediabetes), these data suggest that exposure to cadmium precedes the development of diabetes. In addition, we observed a small association between smoking, as measured by serum cotinine, and the combined outcomes (IFG + diabetes). This finding is consistent with several (but not all) prospective studies of smoking and type 2 diabetes (17) and with the results of a large cross-sectional study that found a significant association between smoking and glycosylated hemoglobin (18).

In rats and mice, cadmium damages pancreatic β-cells, reduces glucose tolerance, and is diabetogenic (19–22). Our analyses suggest that cadmium may have similar effects in humans. However, it is important to emphasize that these results were obtained in a cross-sectional study in which exposure and disease were measured at the same time. A prospective study, in which cadmium levels are determined before the development of disease, will be required to establish a causal basis for these associations.

Supported by RO3 CA 89798 to G.G.S.; D.I. was supported in part by CA57707-07.

We thank Dr. Lynne Wagenknecht for reviewing this manuscript.