Metals in Urine and Peripheral Arterial Disease

1Peripheral arterial disease (PAD) is a consequence of atherosclerotic occlusion of blood flow in the muscular arteries of the lower extremities. Because traditional risk factors for atherosclerosis, such as age, smoking, hypertension, and diabetes, do not completely explain the distribution of PAD in the population, there is considerable interest in identifying novel, nontraditional risk factors. Certain metals may promote atherosclerosis by increasing oxidative stress [e.g., by catalyzing the production of reactive oxygen species or inhibiting their degradation] or by affecting other cardiovascular risk factors [e.g., by increasing blood pressure levels]. Arsenic, for instance, has been associated with PAD and with other forms of atherosclerosis in populations exposed to arsenic in drinking water. However, evidence for an association between most metals and PAD is limited.

2Blood cadmium and lead, at levels well below current safety standards, were associated with an increased prevalence of PAD in the 1999–2000 National Health and Nutrition Examination Survey (NHANES). Cadmium and lead have been associated with other cardiovascular end points, such as myocardial infarction and stroke, in some studies but not all. For several other metals, including barium, cobalt, molybdenum, antimony, and thallium, data on their role in human atherosclerosis are sparse. There are no epidemiologic data on the potential cardiovascular effects of cesium or tungsten.

3The objective of this study was to evaluate the association of metal levels in urine with the prevalence of PAD in NHANES 1999–2000, a representative sample of the civilian, noninstitutionalized U.S. population. In epidemiologic studies, PAD is usually defined by the presence of a low (< 0.9) blood pressure ankle-brachial index (ABI)—that is, when the systolic blood pressure measured at the ankle is < 90% of the systolic blood pressure measured at the arm. We hypothesized that cadmium and lead would be positively associated with PAD, as determined by an ABI < 0.9. In addition, we explored the association of PAD with urinary barium, cobalt, cesium, molybdenum, antimony, thallium, and tungsten, without prior hypotheses. These metals were selected for study in NHANES 1999–2000 by the National Center for Health Statistics (NCHS). The widespread exposure to these metals in the general population and the lack of relevant population data for most of them underscore the public health relevance of this study.

4The NCHS has conducted a series of cross-sectional health and nutrition surveys beginning in the 1960s. These surveys have used a stratified, multistage probability cluster design to provide data representing the noninstitutionalized U.S. population. In 1999, NHANES became a continuous survey and also began collecting information on ABI in men and women ≥40 years of age.

5Among the 9,965 participants in NHANES 1999–2000, one-third of the study population ≥6 years of age were randomly selected to obtain urinary measurements of metals (n = 2,465). NHANES 1999–2000 included a detailed lower-extremity examination that involved ABI measurements in subjects ≥40 years of age with no bilateral amputations and weighing < 400 lb (n = 2,875, of whom 796 had a determination of urinary metals). From these 796 participants, we excluded one participant with an ABI value > 1.5 [usually related to noncompressible vessels in the legs] and five participants with no information on smoking status or urine levels of creatinine, for a final sample size of 790 participants. A few remaining participants had missing data for some of the metals (sample size available for each metal shown in Table 1). The response rate for the household interview and the physical exam components of NHANES 1999–2000 were 82 and 76%, respectively.

6A specific protocol was used to measure ABI in NHANES 1999–2000. The measurements of blood pressure used for ABI were different from the measurements of blood pressure used to define hypertension. Supine systolic blood pressure was measured on the right arm (brachial artery) and both ankles (posterior tibial arteries) using a Doppler device, the Parks Mini-Lab IV, model 3100 (Parks Medical Electronics, Aloha, OR, USA). If the participant had a condition that would interfere with blood pressure reading in the right arm, the left arm was used. Systolic blood pressure was measured twice at each site for participants 40–59 years of age and once at each site for participants ≥60 years of age. The measurements for left and right ABI were obtained by dividing the mean ankle systolic blood pressure in each side by the mean brachial systolic blood pressure. PAD was defined as an ABI value < 0.90 in at least one leg.

7A casual (or spot) urine specimen was collected from the participants after confirmation of no background contamination in collection materials. Urinary levels of cadmium, lead, barium, beryllium, cobalt, cesium, molybdenum, platinum, antimony, thallium, and tungsten were measured at the Environmental Health Sciences Laboratory of the Centers for Disease Control and Prevention (CDC) and the National Center for Environmental Health (NCEH) by inductively coupled plasma–mass spectrometry (PerkinElmer/SCIEX model 500; PerkinElmer, Shelton, CT, USA) using a multielement analytical technique following published protocols. Cadmium levels in urine were corrected for interference from molybdenum oxide. Urine Standard Reference Material 2670 from the National Institute of Standards and Technology (Gaithersburg, MD, USA) was used for external calibration and spiked pools prepared at the laboratory were used for internal quality control. Quality control samples included both bench and blind samples. The ranges for the interassay coefficients of variation for each metal are shown in Table 1. The limits of detection varied by metal, from 0.01 μg/L for thallium to 0.85 μg/L for molybdenum (Table 1). The levels of beryllium and platinum were below the limit of detection in > 98% of participants and were not considered further. For the other metals, the percentage of subjects with levels below the limit of detection ranged from 0.5% for cesium to 30% for tungsten (Table 1). A level equal to the limit of detection divided by the square root of 2 was imputed for those subjects with levels below the limits of detection.

8Information on age, sex, race/ethnicity, education, and smoking was based on self-report. Urinary creatinine was determined using an enzymatic reaction measured with a CX3 analyzer.

9All statistical analyses were performed using SUDAAN software (Research Triangle Institute, Research Triangle Park, NC, USA) to account for the complex sampling design and weights in the NHANES 1999–2000 metal subsample. The jackknife “leave-one-out” method was used to obtain appropriate standard errors of all estimates.

10Urinary metal levels were right skewed and were log-transformed to improve normality. For each metal, the ratio of the geometric mean urinary level and its 95% confidence interval (CI) comparing subjects with PAD versus subjects without PAD were estimated using linear regression models on log-transformed metal levels. For risk analyses, logistic regression was used to obtain the adjusted odds ratios and 95% CIs of PAD comparing the 75th with the 25th percentile of the distribution of each metal assuming a log-linear relationship between urinary metals and PAD. We also used restricted cubic spline transformations to assess nonlinear relationships. All models were adjusted for age, sex, race, and education level. In addition, we evaluated the impact of further adjusting for smoking status, a source of cadmium, lead, and cobalt. Metal levels were based in one casual urine sample and thus depend on urine dilution. Urinary metal levels were adjusted for urinary creatinine to correct for differences in urine dilution in spot urine samples, but because of the limitations of creatinine concentration to adjust for urine dilution, we present results both with and without creatinine adjustment. We also repeated the analyses excluding subjects fasting < 8 hr to evaluate the impact of postprandial or fasting state and rapid renal clearance of some metals. No substantial changes were observed (data not shown). We also assessed confounding by other cardiovascular risk factors, including hypertension, hypercholesterolemia, diabetes, estimated glomerular filtration rate, and C-reactive protein, by adding them one at a time to the multivariable models. No noticeable changes were observed (data not shown), and because of the limited sample size, these additional variables were not included in the final models.

11Table 1 describes urinary metal levels in the study sample. The geometric means were lowest for tungsten, 0.07 μg/L, and highest for molybdenum, 37.7 μg/L. The median level and the interquartile range for each metal by participant characteristics are shown in Figure 1. Men had higher levels than women for most metals, and cadmium and lead tended to increase with age. Compared with whites, blacks and Mexican Americans had higher levels of cadmium, lead, and tungsten, and blacks had higher levels of cesium, molybdenum, antimony, and thallium. Current smokers had higher levels of cadmium, lead, antimony, and tungsten compared with never smokers.

12The overall prevalence of PAD in the study sample was 5.4% (54 cases). After adjusting for demographic variables, smoking status, and creatinine in urine, subjects with PAD had 36% (95% CI, 1 to 83) higher mean levels of cadmium in urine compared with subjects without PAD (Table 2). Levels of lead, barium, cobalt, cesium, molybdenum, antimony, and thallium were similar in subjects with and without PAD. Subjects with PAD had 49% (95% CI, −10 to 249) higher mean levels of tungsten. The association of cadmium and tungsten with PAD was also evident in logistic models (Table 3, Figure 2). The odds ratio for PAD comparing the 75th with the 25th percentile of the cadmium distribution was 3.05 (95% CI, 0.97 to 9.58). The corresponding odds ratio for tungsten was 2.25 (95% CI, 0.97 to 5.24). For antimony, no association was apparent in linear models (Tables 2 and 3), but nonlinear models showed an increase in the prevalence of PAD at very low levels and a plateau of risk > 0.1 μg/L (Figure 2).

13Cadmium levels in urine were 36% higher in subjects with PAD than in those without PAD. This association was similar but actually stronger than the previously reported association between blood cadmium and PAD in NHANES 1999–2000, where blood cadmium was 16% higher in PAD cases than in noncases. The stronger association for urinary compared with blood cadmium probably reflects the fact that urinary cadmium is a more reliable biomarker of chronic cadmium exposure than blood cadmium. The association between cadmium levels and PAD was not explained by smoking status, and adjusting for smoking decreased only slightly the magnitude of the association. In our study, an increased prevalence of PAD was associated with low levels of urinary cadmium, below the levels reported in workers exposed to cadmium occupationally. For example, only two subjects had cadmium levels > 3 μg/g of creatinine, the Occupational Safety and Health Administration safety standard for cadmium in urine. In the NHANES population, because occupational exposure is expected to be relatively rare, it is likely that exposure to cadmium occurred mainly through cigarette smoking, inhalation of airborne cadmium in ambient air (usually higher near coal-fired power plants and municipal waste incinerators), or consumption of some foods (highest levels in shellfish, liver, and kidney meats).

14Several mechanisms may explain an increased risk of atherosclerosis with cadmium, including the catalysis of reactive oxygen species, the promotion of lipid peroxidation, the depletion of glutathione and protein-bound sulf-hydryl groups, the production of inflammatory cytokines, and the down-regulation of nitric oxide production. Cadmium has also induced atherosclerosis and hypertension in some animal models in vivo. However, these mechanistic studies were typically conducted at considerably higher exposures than those corresponding to the urinary concentrations observed in the present study, and hence their relevance to human atherogenesis and to PAD is uncertain.

15Epidemiologic studies of cadmium and cardiovascular disease are also limited. Ecologic studies have found associations of cardiovascular mortality rates with cadmium levels in air and in soil and water. Two small case–control studies found higher blood cadmium in subjects with myocardial infarction than in controls, but a cross-sectional study in Belgium found no association between blood cadmium and the prevalence of cardiovascular disease. Finally, several autopsy studies have found associations between tissue cadmium levels and atherosclerotic lesions. Additional studies, particularly with a prospective design, are needed to confirm the role of cadmium in the pathogenesis of PAD and to determine its role in other cardiovascular end points.

16Lead in urine was not associated with PAD in this study. This result is in contrast to the association observed between blood lead and PAD in NHANES 1999–2000, where blood lead levels were 14% higher in cases of PAD than in noncases. This discrepancy could be related to the fact that urinary lead levels in this population were low and may have considerable fluctuations when evaluated in spot urine samples. Under these conditions, urinary lead is considered a less reliable biomarker of exposure than blood lead. Indeed, previous cohort studies of lead and cardiovascular disease have used blood lead as the biomarker of exposure. Blood lead was positively associated with cardiovascular mortality in NHANES II and with coronary heart disease incidence in Denmark, although an earlier study in British men did not show an association between blood lead and cardiovascular disease incidence. Several mechanisms support a role for lead in atherosclerosis. Lead increases blood pressure, and experimental studies show that lead promotes oxidative stress, stimulates inflammation, and induces endothelial damage. The role of lead in the development of atherosclerosis, however, needs to be further investigated in mechanistic studies at low levels of exposure and in prospective studies in humans using appropriate biomarkers of chronic exposure.

17Tungsten levels in urine were 49% higher in subjects with PAD than in those without PAD. This analysis is among the first to examine the role of tungsten in any health status indicator, and it needs to be interpreted cautiously in the context of multiple comparisons. Little is known regarding tungsten toxicity and carcinogenicity, and there are insufficient data, either from animal or human studies, on its cardiovascular effects. It is known, however, that tungsten is thrombogenic and proinflammatory. In fact, these properties have motivated the clinical use of tungsten coils for the occlusion of intracranial aneurysms, varicocele veins, and other abnormal vascular connections. In addition, tungsten may interfere with the biologic effect of molybdenum, an essential trace element that acts as cofactor in several proteins. For instance, tungsten inhibits xanthine oxidase, an antioxidant molybdoenzyme with a role in endothelial dysfunction and in the maintenance of the vessel wall integrity.

18Sources of tungsten exposure include occupational use of tungsten inert gas, welding, and sometimes drinking water. The amounts of tungsten in foods and in ambient air are generally unknown. Urban settings tend to have higher levels of tungsten in the air because tungsten can be released from industrial sources and waste incinerators (tungsten filaments are used in incandescent light bulbs). In response to the lack of knowledge on the health effects of tungsten exposure coupled with evidence of widespread human exposure, the CDC/NCEH nominated tungsten in 2002 to the National Toxicology Program as a priority candidate for toxicologic assessment. Our findings suggest that these studies should also include the evaluation of the relationship of tungsten to cardiovascular end points and its potential mechanisms of action.

19For antimony, there was an increase in the prevalence of PAD in subjects with low levels compared with those at the limit of detection, and the risk remained increased > 0.1 μg/L. The general population is exposed to antimony in food, in drinking water, and in ambient air. At levels found in antimony smelting plants, antimony has been related to pneumoconiosis and dermatitis. Although antimony is a well-known toxic metal at high doses, there is only one available study of chronic antimony exposure and cardiovascular end points in humans. This study compared coronary mortality in Mexican-American antimony smelter workers with other groups of workers, but the findings were inconclusive. Interestingly, antimony shares similar chemical and toxicologic properties with arsenic, and they are frequent coexposures. Arsenic exposure has been associated with PAD, and “blackfoot disease” is a classic sign of high arsenic exposure. Unfortunately, arsenic levels were not measured in NHANES 1999–2000, and we cannot discard the possibility that the observed association between PAD and antimony was due to arsenic coexposure. Further studies are required on antimony and cardiovascular end points.

20Barium, cobalt, cesium, molybdenum, and thallium in urine were not associated with PAD in this representative sample of the U.S. population. Data on the possible role of these metals in atherosclerosis are scarce. Ecologic studies have found positive associations with cardiovascular disease for barium and thallium and a negative association for molybdenum. Thallium is a poison at high dose. At low dose, thallium is used in cardiac imaging and thought to be relatively safe. Cobalt exposure in the hard metal industry results in cardiomiopathy, but no information is available on its atherogenicity.

21Our analysis was limited by a relatively small number of cases, because NHANES 1999–2000 measured metals in urine in just one-third of survey participants. This sample size limited the investigation of interactions (e.g., lead and cadmium, or tungsten and molybdenum) and makes it possible that we missed weak associations between some metals and PAD. At the same time, our prior hypotheses regarding the associations of cadmium and lead with PAD were specific, but for the other metals our analyses were exploratory and need to be confirmed in future studies.

22Other limitations include the cross-sectional design of the study, the possibility of residual confounding by socioeconomic status or urbanization, the use of a single measurement of urinary metals, and the lack of 24-hr urine collection to better account for short-term variability in metal excretion and urine dilution. To correct for urine dilution, we adjusted for urinary creatinine. However, creatinine in urine is a marker of both urine dilution and creatinine production, and it is associated with factors that affect production such as sex, age, race, and muscle mass. The adequacy of correcting for creatinine has been questioned, and in fact the Second National Report on Human Exposure to Environmental Chemicals, based on NHANES data, presented metal concentrations in urine both ways, with and without correction for urinary creatinine. In our data, the findings using models with and without adjustment for creatinine were similar and do not affect the conclusions.

23Despite these limitations, this study is the first systematic investigation of a panel of metals in urine with PAD. The use of a representative sample of the U.S. population, rigorous laboratory methods with extensive quality control, and the availability of a standardized procedure to measure ABI add to the strengths of this study.

24Urinary cadmium was strongly associated with PAD in a representative sample of the general U.S. population. This finding strengthens similar results from a previous study in the same population using blood cadmium as a biomarker of exposure and further supports a possible role for cadmium in atherosclerosis. Tungsten was associated with PAD throughout the range of exposure, whereas antimony showed a positive association at low levels that reached a plateau > 0.1 μg/L. These associations need to be interpreted cautiously and require confirmation in future epidemiologic studies and supporting evidence from mechanistic research. Finally, no association between PAD and other metals in urine (lead, barium, cobalt, cesium, molybdenum, and thallium) was evident at the levels found in the general population. The observation of widespread exposure, particularly to poorly studied metals such as tungsten and antimony, supports further research on the role of metals and PAD.