Urine Arsenic Concentrations and Species Excretion Patterns in American Indian Communities Over a 10-year Period: The Strong Heart Study

1Inorganic arsenic (arsenite, arsenate) is a naturally occurring toxicant and carcinogen that contaminates groundwater supply systems in countries around the world. In the United States, arsenic levels in drinking water > 10 μg/L—the U.S. Environmental Protection Agency (EPA) maximum contaminant level—disproportionately affect small communities in the West, Midwest, and Northeast regions. Flour and rice also contain inorganic arsenic, particularly if grown or cooked in areas with arsenic contamination in soil and water. The metabolism of inorganic arsenic in the human body results in methylarsonate (MA) and dimethylarsinate (DMA), which are excreted in urine together with unchanged inorganic arsenic. Seafood is a source of organic arsenic compounds (arsenobetaine, arsenosugars, arsenolipids) that have no or low toxicity compared with inorganic arsenic.

2In populations with low seafood intake, total urine arsenic and the sum of inorganic arsenic and methylated (MA and DMA) urine arsenic species are established biomarkers that integrate inorganic arsenic exposure from multiple sources. The proportion of arsenic species that is excreted as inorganic arsenic, MA, or DMA also provides information on the metabolism of inorganic arsenic in the human body. The arsenic species excretion pattern in human urine is approximately 10–20% inorganic arsenic, 10–20% MA, and 60–80% DMA, with substantial variation among individuals. Individual arsenic excretion patterns, on the other hand, were fairly constant over time in studies of up to 1 year of follow-up. Because a higher proportion of MA in urine has been associated with an increased risk of cancer and cardiovascular outcomes, there is substantial interest in characterizing the long-term arsenic species excretion patterns, especially for MA.

3The objective of this study was to conduct an initial assessment of arsenic exposure and excretion pattern, as measured by total urine arsenic and urine arsenic species, in American Indians from Arizona, Oklahoma, and North and South Dakota who participated in the Strong Heart Study, a population-based prospective cohort study funded by the National Heart, Lung, and Blood Institute. On the basis of arsenic concentrations measured in public drinking water systems in the Strong Heart Study communities, we expected arsenic exposure levels to be higher in Arizona, intermediate in the Dakotas, and lower in Oklahoma. Given dietary patterns in the study communities, we expected arsenobetaine concentrations, a marker of seafood arsenicals, to be very low. Using three urine samples for each participant, we also assessed arsenic exposure and excretion patterns over a 10-year period and the feasibility of measuring arsenic species in long-term stored urine samples frozen at − 70°C.

4The Strong Heart Study was designed to investigate cardiovascular disease and its risk factors in 13 American Indian communities located in Arizona, Oklahoma, and North and South Dakota. From 1989 to 1991, enrolled tribal members 45–74 years of age who were living in those communities were invited to participate. The aim was to recruit approximately 1,500 participants per region. In Arizona and Oklahoma, every eligible person was invited to participate, in person or by letter. In North and South Dakota, a cluster sampling technique was used. A total of 4,549 participants were recruited in 1989–1991 for a baseline visit and invited to participate in subsequent visits in 1993–1994 and 1998–1999. The participation rates were 62% at baseline and 89% and 88% for surviving participants at visits 2 and 3, respectively. All study visits included a questionnaire, a physical examination, and biological specimen collection. The Strong Heart Study protocol and consent form were approved by the Institutional Indian Health Service Review Boards and by the participating Indian communities. Informed consent was obtained from all participants. For the present study, we randomly selected 60 Strong Heart Study participants (20 from Arizona, 20 from Oklahoma, and 20 from the Dakotas) among those for whom we had stored urine samples from the three study visits (n = 3,197). Participants’ characteristics (age, sex, body mass index, smoking, and alcohol use) in the present study were comparable to the overall study population (Table 1).

5Spot urine samples were collected in polypropylene tubes, frozen within 1–2 hr of collection, shipped in dry ice, and stored from 8 to 18 years at − 70°C in the Penn Medical Laboratory, MedStar Research Institute (Washington, DC, USA). The freezers have been operating under a strict quality control system to guarantee secure sample storage. For arsenic analyses, urine samples were thawed in August 2007, and up to 1.0 mL was transferred to a small vial, transported on dry ice to the Trace Element Laboratory, Graz University, Austria, and stored at − 80°C until analysis. Urine samples were frozen for an average of 13 years (range, 8–18 years) before analysis.

6We measured total urine arsenic concentrations (expressed on an elemental basis) using inductively coupled plasma/mass spectrometry (ICPMS). The limit of quantification for total urine arsenic was 0.1 μg/L. We checked measurement accuracy with 1+19 diluted human urine no. 18 from Japan’s National Institute of Environmental Studies (Tsukuba, Japan) (e.g., 0.5 mL urine + 9.5 mL water). The measured mean (± SD) total arsenic concentration of 144 ± 4 μg/L (n = 19) was in agreement with the certified concentration of 137 ± 11 μg/L. Total arsenic concentrations exceeded the limit of quantification in all samples.

7Urine concentrations of arsenite, arsenate, MA, and DMA (expressed on an elemental basis) were measured using high-performance liquid chromatography/vapor generation ICPMS. The limits of quantification were 0.1 μg/L for arsenite and 0.5 μg/L for arsenate, MA, and DMA. Arsenite and arsenate were below the limit of quantification in 2 (1%) and 126 (70%) samples, respectively. MA and DMA exceeded the limit of quantification in all samples. The interassay coefficients of variation for an in-house reference urine sample for arsenite, arsenate, MA, and DMA were 3.8%, 4.5%, 4.3%, and 1.9%, respectively (n = 18). We did not detect thio-DMA, an arsenic species that has been related to arsenosugar exposure and to high arsenic exposure in Bangladesh, in any of the study samples.

8We measured urine arsenobetaine concentrations using cation-exchange chromatography on a Zorbax 300 SCX column (4.6 mm inner diameter × 250 mm; Agilent, Waldbronn, Germany) operated at 30°C. The mobile phase was 10 mM pyridine (pH 2.3, adjusted with formic acid) at a flow rate of 1.5 mL/min. We injected 20 μL of sample. The limit of quantification for urine arsenobetaine was 0.5 μg/L. We checked the accuracy of the measurements with 1+9 diluted human urine no. 18. The mean (± SD) measured value for arsenobetaine was 68 ± 2 μg/L (n = 18), in agreement with the certified concentration of 69 ± 12. Ninety-six (53%) samples had concentrations below the limit of quantification, reflecting infrequent seafood intake in the study population. The median (10th–90th percentiles) of urine arsenobetaine was 0.5 μg/L creatinine (< 0.5–137) [0.5 μg/g (< 0.5–6.1)]. Urine arsenobetaine concentrations were similarly low by region and other participant characteristics (data not shown).

9Baseline information on sociodemographic data (age, sex, study region), smoking history, alcohol use, and body mass index (kilograms per square meter) were obtained from the Strong Heart Study questionnaires and physical examinations. Urine creatinine was measured at the laboratory of the National Institute of Diabetes and Digestive and Kidney Diseases Epidemiology and Clinical Research Branch (Phoenix, AZ, USA) by an automated alkaline picrate methodology run on a rapid flow analyzer. To account for urine dilution in spot urine samples, we divided urine arsenic (micrograms per liter) by urine creatinine (grams per liter) and expressed the concentrations of total urine arsenic and its species in terms of micrograms per gram creatinine.

10Because the limit of quantification was different for arsenite and arsenate and 70% of participants had arsenate concentrations below the limit of quantification, we used a bivariate Tobit multilevel model to impute these values in participants with arsenite or arsenate levels below the limit of quantification. The bivariate Tobit multilevel model had random participant and region-specific intercepts and used total urine arsenic concentrations as an explanatory variable. Arsenite and arsenate concentrations in participants with values below the limit of quantification were imputed as the median of the participant-specific posterior distribution derived from the bivariate Tobit multilevel model. For arsenite, the posterior distribution median was 0.08 μg/L for each of the two samples below the limit of quantification. For arsenate, the median (10th–90th percentiles) of the posterior distribution medians for the 126 urine samples below the limit of quantification was 0.13 μg/L (0.07–0.26). We calculated urine inorganic arsenic concentration as the sum of arsenite and arsenate concentrations.

11Because urine concentrations of total arsenic and its species were right-skewed, we used medians and 10th and 90th percentiles to describe their distributions. To evaluate variation in arsenic excretion patterns, we calculated the percentages of inorganic arsenic, MA, and DMA over the sum of inorganic and methylated species (%iAs, %MA, and %DMA). These percentages were normally distributed; we describe them using means ± SDs.

12To evaluate the determinants of biomarkers of inorganic arsenic exposure at baseline and their trajectories over time, we used mixed linear models for change for log-transformed total arsenic, inorganic arsenic, MA, and DMA concentrations in urine. We used study region, sex, age, body mass index, smoking status, and alcohol consumption as fixed effects (explanatory variables). From these models, we estimated the ratios of the geometric means of baseline urine arsenic concentrations and of their annual rates of change comparing different categories of the explanatory variables. In sensitivity analyses, we also ran generalized estimated equation (GEE) models, overall and stratified by region. Findings were similar (data not shown).

13We also used mixed linear models for change to model the %MA over the sum of inorganic and methylated arsenic species and to estimate the mean differences in baseline %MA and in its annual rate of change comparing different categories of study region, sex, age, body mass index, smoking status, and alcohol use. In sensitivity analyses, we also ran GEE models, overall and stratified by region. Findings were similar (data not shown).

14We calculated intraclass correlation coefficients (ICCs) to describe the extent of within-individual compared with between-individual variability over time. A high ICC (close to 1) indicates that the within-individual variability is relatively small compared with the between-individual variability and that biomarkers of arsenic exposure and metabolism are relatively stable for an individual compared with between-individual variability. We performed all statistical analyses using the program R (R Project for Statistical Computing, Vienna, Austria).

15The mean (± SD) age of study participants was 54 ± 8 years, and 43% were men, with small differences across regions (Table 1). Similar to the overall Strong Heart Study population, participants from Arizona had higher body mass index compared with participants from Oklahoma and the Dakotas, and participants from the Dakotas were more likely to be current smokers and to use alcohol compared with participants from Arizona and Oklahoma.

16Median (10th–90th percentiles) concentrations for total arsenic and for the sum of inorganic and methylated arsenic species at baseline (visit 1) were 10.5 μg/g creatinine (4.0–27.8) [12.5 μg/L (4.3–42.7)] and 7.2 μg/g (3.1–16.9) [9.8 μg/L (2.8–28.9)], respectively (Table 2). By region (Table 3, Figure 1), the median sum of inorganic and methylated arsenic species was higher in Arizona (12.5 μg/g), intermediate in the Dakotas (9.1 μg/g), and lower in Oklahoma (4.4 μg/g). After adjustment for sex, age, body mass index, smoking, and alcohol use, participants in Oklahoma and the Dakotas had approximately 80% and 29% lower urine concentrations, respectively, for the sum of inorganic and methylated arsenic species compared with participants in Arizona (Table 4), although the differences were statistically significant only for Oklahoma compared with Arizona. After multivariable adjustment, study region was the only statistically significant determinant of the sum of inorganic and methylated species at baseline (Table 4).

17The sum of inorganic and methylated arsenic species remained relatively similar over 10 years of follow-up (Table 2, Figure 1). The ICC for the sum of inorganic and methylated species was 0.80, with little difference across regions (Figure 1). After multivariable adjustment, the average annual ratio over the 10-year follow-up was 1.23 [95% confidence interval (CI), 0.92–1.64], with no difference by participant characteristics (data not shown).

18The means (± SDs) for %iAs, %MA, and %DMA over the sum of inorganic and methylated species at baseline (visit 1) were 10.6 ± 5.2%, 18.4 ± 5.4%, and 70.9 ± 10.0%, respectively (Table 1). By region (Table 3, Figure 2), the mean %MA over the sum of inorganic and methylated arsenic species was lower in Oklahoma (15.4%) compared with Arizona (19.4%) and the Dakotas (20.5%). After multivariable adjustment, mean %MA was 6.0% (95% CI, 3.8–8.2%) point increased in men compared with women, and − 3.4% (95% CI, − 6.1 to − 0.7%) point decreased in participants in Oklahoma compared with Arizona (Table 5).

19The percentage distribution of arsenic species remained relatively similar over 10 years of follow-up (Table 2, Figure 2). The overall ICCs for %iAs, %MA, and %DMA were 0.64, 0.80, and 0.77, respectively, with little difference across regions. After multivariable adjustment, the average annual change over the 10-year follow-up in %MA over the sum of inorganic and methylated species was − 0.7% (95% CI, − 1.5 to 0.2%), with no difference by participant characteristic (data not shown).

20This study quantified for the first time individual biomarkers of arsenic exposure and metabolism, as assessed in urine, in American Indian populations from Arizona, Oklahoma, and the Dakotas over a 10-year period. Total urine arsenic and the sum of inorganic and methylated arsenic species concentrations were higher in Arizona, intermediate (but high in some samples) in the Dakotas, and lower in Oklahoma, as expected based on data on arsenic in groundwater. For the sum of inorganic and methylated compounds, four samples (two in Arizona and two in the Dakotas) exceeded 35 μg/g creatinine, a safety standard used to monitor inorganic arsenic exposure in occupational settings. Arsenobetaine concentrations, related to seafood intake, were low compared with those in the U.S. general population, reflecting low seafood intake in the study communities. The constancy in urine arsenic concentrations in three urine samples collected over a 10-year period (1989–1999) is consistent with previous studies measuring arsenic in private and public drinking water systems over long periods of time and support the observation that arsenic levels in drinking water are stable in the absence of public health interventions or changes in water sources. Our data also demonstrate the feasibility of using long-term stored urine samples for determining arsenic species in population-based studies.

21The three Strong Heart Study visits were conducted between 1989 and 1999, before the U.S. EPA established the new arsenic maximum contaminant level in drinking water (10 μg/L). Data from public water systems in the states where the Strong Heart Study was conducted show that arsenic was higher in Arizona (county-level 95th percentile, 20.5 μg/L) compared with Oklahoma (county-level 95th percentile, 2.0 μg/L) and North and South Dakota (county-level 95th percentile, 4.0 μg/L). In the Strong Heart Study communities, arsenic concentrations measured in public drinking water systems in the 1990s and 2000s ranged from < 10 to 61 μg/L in Arizona and from < 1 to > 21 μg/L in North and South Dakota (no data are available for the Oklahoma communities, but levels were expected to be < 10 μg/L). Many participants in the Strong Heart Study rely upon private wells for water. Private wells do not need to comply with the U.S. EPA arsenic standard, and the water arsenic concentrations in many of them exceed 10 μg/L, particularly in the western states.

22Urine arsenic concentrations in participants from Oklahoma were comparable with those for a representative sample of U.S. adults (median total urine arsenic in participants with arsenobetaine below the limit of detection, 3.4 μg/L), reflecting generally low water arsenic concentrations (< 10 μg/L) in the United States. Urine arsenic concentrations for participants from the Dakotas and in particular from Arizona were substantially higher, comparable to urine concentrations in populations in Nevada, California, and Central Europe exposed to moderately high arsenic in drinking water (~ 10 to > 50 μg/L) but much lower than urine concentrations in populations in Taiwan, Chile, or Bangladesh, where drinking-water arsenic concentrations often exceeds 100 μg/L.

23Because arsenic metabolism is thought to play a major role in inorganic arsenic toxicity, there is substantial interest in understanding the distribution and main determinants of arsenic species in urine. The distribution and interindividual variability of urine arsenic species in Strong Heart Study participants were comparable with those reported in other human populations. Studies conducted in Argentina and Chile previously showed a fairly constant within-person urine arsenic excretion pattern up to 1 year of follow-up. In our study, we extend this observation to 10 years, providing additional support for genetic factors influencing interindividual differences in arsenic metabolism. Contrary to our study in the United States and previous studies in Argentina and Chile, a recent study from Bangladesh measured low within-person reproducibility for %MA over a 2-year period.

24Similar to studies in the United States, Europe, Chile, Taiwan, and Bangladesh, men in our study had higher %MA in urine than did women. Some studies have also reported unadjusted associations of smoking and alcohol with higher %MA in urine. In our study, after adjustment for sex, age, region, and body mass index, smoking and alcohol were not statistically associated with higher %MA, although our sample size was small. After adjustment for sex and other variables, participants in the Dakotas had a lower %MA concentration in urine compared with participants in Arizona or Oklahoma. The genetic determinants and health implications of differences in urine arsenic excretion patterns across study regions require further investigation.

25Strengths of this study include the availability of repeated urine samples and the quality of the study measures. The Strong Heart Study used a standardized protocol for the recruitment of study participants, interviews, physical examinations, and collection and storage of urine samples. Some limitations include the relatively small sample size, the use of spot urine samples, the relatively high limit of quantification for arsenate, the storage of urine samples for > 10 years, and the difficulty in directly evaluating the impact of time of freezer storage. The long-term stability of inorganic arsenic, MA, DMA, and arsenobetaine in stored urine samples has been evaluated for a 6- to 8-month period and is further supported by the findings of our study. Arsenic species such as methylarsonite [MA(III)] and dimethylarsinite [DMA(III)], however, tend to quickly convert to their corresponding pentavalent forms (MA and DMA) and could not be measured. Arsenite and arsenate generally interconvert in urine depending on handling and storage conditions, and the sum of the two is generally used to measure inorganic arsenic in urine. In our study, to deal with data missing due to a relatively high limit of quantification for arsenate, we imputed arsenite and arsenate concentrations below the limit of quantification using total arsenic as an explanatory variable. In an alternative approach, all arsenate could have been reduced to arsenite (or all arsenite could have been oxidized to arsenate) before laboratory analysis to provide a stable measure of inorganic arsenic in urine. Overall, although arsenic-related health outcomes could not be assessed because of the limited sample size, our study supports the feasibility of using long-term stored urine samples for measuring arsenic exposure and metabolism in population-based studies such as the Strong Heart Study.

26Arsenic in drinking water continues to be a public health problem in the United States, affecting several millions of Americans who live in small communities and rural areas, partly because private wells are not covered by the U.S. EPA maximum contaminant level, but also because of insufficient financial resources and public health infrastructure to ensure compliance with the arsenic standard in small communities. The well-established carcinogenic effect of arsenic, and evidence indicating a potential role of arsenic on cardiovascular disease and diabetes highlight the importance of ensuring arsenic levels below the U.S. EPA maximum contaminant level in all water supply systems, including those serving small communities.

27This study found low to moderate inorganic arsenic exposure and long-term constancy in urine arsenic concentrations and excretion patterns among American Indians from Arizona, Oklahoma, and the Dakotas who participated in a large population-based prospective cohort study. In communities with stable environmental exposure to inorganic arsenic and with relatively low fish intake, our study showed that a single determination of urine arsenic and its species provides a useful biomarker of ongoing arsenic exposure and of individual arsenic metabolism. Additional research is needed to further characterize the genetic and environmental determinants of arsenic metabolism and the health impact of arsenic exposure and metabolism, including cardiovascular disease and diabetes, in populations exposed to low and moderate levels of inorganic arsenic in drinking water.