Arsenic Exposure From Drinking Water, Arsenic Methylation Capacity, and Carotid Intima-Media Thickness in Bangladesh

Abstract

We conducted a cross-sectional study to evaluate the interrelationships between past arsenic exposure, biomarkers specific for susceptibility to arsenic exposure, and carotid intima-media thickness (cIMT) in 959 subjects from the Health Effects of Arsenic Longitudinal Study in Bangladesh. We measured cIMT levels on average 7.2 years after baseline during 2010–2011. Arsenic exposure was measured in well water at baseline and in urine samples collected at baseline and during follow-up. Every 1-standard-deviation increase in urinary arsenic (357.9 µg/g creatinine) and well-water arsenic (102.0 µg/L) concentration was related to a 11.7-µm (95% confidence interval (CI): 1.8, 21.6) and 5.1-µm (95% CI: −0.2, 10.3) increase in cIMT, respectively. For every 10% increase in monomethylarsonic acid (MMA) percentage, there was an increase of 12.1 µm (95% CI: 0.4, 23.8) in cIMT. Among participants with a higher urinary MMA percentage, a higher ratio of urinary MMA to inorganic arsenic, and a lower ratio of dimethylarsinic acid to MMA, the association between well-water arsenic and cIMT was stronger. The findings indicate an effect of past long-term arsenic exposure on cIMT, which may be potentiated by suboptimal or incomplete arsenic methylation capacity. Future prospective studies are needed to confirm the association between arsenic methylation capacity and atherosclerosis-related outcomes.

Inorganic arsenic (InAs) occurs naturally in groundwater in many parts of the world, affecting millions of people worldwide. Chronic exposure to InAs from drinking water has been associated with increased risks of preclinical and clinical cardiovascular disease endpoints (1–8). However, the underlying mechanism is not clear.

The primary metabolic pathway of InAs in humans is methylation. Once ingested, InAs is methylated to monomethylarsonic acid (MMA) and dimethylarsinic acid (DMA). The relative distribution of arsenic metabolites in urine varies from person to person and has been interpreted to reflect arsenic methylation capacity (9, 10). In particular, MMA is believed to be more toxic among these metabolites, and incomplete methylation, indicated by a high percentage of MMA (MMA%) in urine, has been consistently related to cancer risk (11–14). However, the relationship between urinary MMA% and preclinical or clinical endpoints for cardiovascular disease is unknown. Evaluation of the relationship between urinary MMA% and cardiovascular disease-related endpoints may help to identify susceptibility and increase understanding of the exposure-disease mechanisms.

Carotid intima-media thickness (cIMT) is an ultrasonographic measure of early atherosclerosis and a valid surrogate marker for clinical endpoints (15, 16). Experimental studies have suggested that arsenic exposure may induce atherosclerosis by induction of oxidative stress, inflammatory responses, and endothelial dysfunction (1). Arsenic may also enhance platelet aggregation and reduce fibrinolysis (17). Several studies in Taiwan found that past arsenic exposure was positively associated with carotid atherosclerosis, defined as cIMT ≥1 mm, with exposure having been measured several decades previously using either the median well-water arsenic concentration at the village level or the arsenic level in household wells (18–24). However, few studies have used biomarkers of arsenic exposure or biomarkers specific to susceptibility to InAs exposure.

Arsenic contamination of drinking water in Bangladesh has been recognized as a massive public health hazard (25). In 2000, we established a large prospective cohort study to evaluate the health effects of arsenic exposure. In the current analysis, we evaluated the associations between baseline arsenic exposure and cIMT measured on average 7.2 years after baseline among 959 subjects randomly selected from the overall cohort. In a subset of 782 participants for whom urinary arsenic metabolites had been measured, we also assessed the relationship between arsenic methylation capacity and cIMT.

MATERIALS AND METHODS

Study population

The parent study, the Health Effects of Arsenic Longitudinal Study (HEALS), is an ongoing prospective cohort study in the Araihazar subdistrict of Dhaka, Bangladesh. A detailed description of the study has been presented elsewhere (26). Briefly, between October 2000 and May 2002, a total of 11,746 men and women aged 18 years or more (the original cohort) were recruited from a well-defined 25-km² geographical area. During 2006–2008, HEALS was expanded to include an additional 8,287 participants (the expansion cohort) following the same methods. The overall participation rate was 97%. Demographic and lifestyle data were collected using a standardized questionnaire. The cohort is being followed up at 2-year intervals with in-person visits, which include a physical examination and collection of urine samples. Informed consent was obtained from the study participants, and the study procedures were approved by the Ethical Committee of the Bangladesh Medical Research Council and the institutional review boards of Columbia University and the University of Chicago.

Arsenic exposure measurements

At baseline, water samples from all 10,971 tube wells in the study area were collected. The samples were acidified to 1% with high-purity Optima hydrochloric acid (Fisher Scientific, Pittsburgh, Pennsylvania) for at least 48 hours before analysis. Total arsenic concentration was analyzed by high-resolution inductively coupled plasma mass spectrometry with a detection limit of <0.2 µg/L. Detailed information on duration and source of exposure has been published elsewhere (27) and is given in the Web Appendix (available at http://aje.oxfordjournals.org/). As one of the eligibility criteria for the study, all participants were required to have been primary users of one of the tested tube wells, designated the “index” well, for at least 3 years. Individuals' choice of well was largely based on geographical convenience, and concentrations of arsenic in well water were not widely known among the study population before recruitment (26–28). In the present study, 92% of the study participants used the index well as their exclusive source of drinking water at baseline. The average duration of index well use was 7.8 years prior to baseline. Among the 958 subjects included in the present study, 8.5% shared a well with 1 other subject and <1% shared a well with 2–3 other subjects. We used the arsenic concentration in the index well assessed at baseline as the baseline level of arsenic exposure from the well.

Spot urine samples were collected at baseline and at all follow-up visits. Total arsenic concentration was measured by graphite furnace atomic absorption, using a PerkinElmer AAnalyst 600 graphite furnace system (PerkinElmer, Waltham, Massachusetts) with a detection limit of 2 μg/L (29). Urinary creatinine level was analyzed using a method based on the Jaffé reaction (30). Urinary arsenic metabolites were measured in baseline urine samples using a method described by Reuter et al. (31). This method uses high-performance liquid chromatography separation of arsenobetaine, arsenocholine, pentavalent InAs (As^V), trivalent InAs (As^III), MMA, and DMA, followed by detection by inductively coupled plasma mass spectrometry with a dynamic reaction cell. Because As^III can oxidize to As^V during sample transport, storage, and preparation, we express them as total InAs (i.e., As^III + As^V). The percentages of InAs, MMA, and DMA (InAs%, MMA%, and DMA%, respectively) in urine were calculated by dividing each metabolite by the sum of InAs, MMA, and DMA. We also constructed 2 methylation indices: the primary methylation index (PMI), namely the ratio of MMA to InAs, and the secondary methylation index (SMI), namely the ratio of DMA to MMA.

Measurement of cIMT

We took cIMT measurements between April 2010 and September 2011 (on average, 7.2 years after baseline). A total of 800 participants were randomly selected from the 11,224 original cohort members who provided urine samples at baseline. A total of 500 participants were randomly sampled from the 5,136 participants over 30 years of age in the expansion cohort. Some participants did not complete cIMT measurements due to death, moving, serious illness, or lack of time. We measured cIMT for 590 persons from the original cohort and 369 persons from the expansion cohort, and the final study population included 959 participants.

Throughout the study, cIMT measurement was conducted using a SonoSite MicroMaxx ultrasound machine (SonoSite, Inc., Bothell, Washington) equipped with an L38e/10-5-MHz transducer by 1 designated physician with extensive training in sonography who was blinded to participants' arsenic exposure levels. We used a specific protocol developed and implemented in the Oral Infections and Vascular Disease Epidemiology Study (INVEST) (32). The carotid arteries were scanned longitudinally in 3 segments, using the lateral extent of each carotid segment as defined relative to the tip of the flow divider, which is normally the most clearly defined anatomical reference in the proximity of the carotid bifurcation. The carotid segments were defined as follows: 1) near and far walls of the segment extending from 10 mm to 20 mm proximal to the tip of the flow divider into the common carotid artery; 2) near and far walls of the carotid bifurcation beginning at the tip of the flow divider and extending 10 mm proximal to the flow divider tip; and 3) near and far walls of the proximal 10 mm of the internal carotid artery. We analyzed cIMT measurements offline with MATLAB (MathWorks, Natick, Massachusetts) software, which automatically calculated the distances between boundaries and expressed the results as the mean and maximal values. The main outcome variable was the mean measurement of the near and far walls of the maximum common carotid artery IMT (mean of 4 sites), similarly to previous studies (16, 33).

Statistical analyses

We modeled cIMT as a continuous variable in all analyses. Multiple linear regression analyses using cIMT as the response variable were conducted to estimate the difference in cIMT associated with a 1-standard-deviation (1-SD) increase in baseline well-water arsenic concentration, a 1-SD increase in baseline total urinary arsenic concentration, 10% increases in baseline urinary InAs%, MMA%, and DMA%, and 1-unit increases in PMI and SMI. Although cIMT was not normally distributed in the study, it was approximately symmetrically distributed. For better interpretation of the results, cIMT values were not transformed in the main analyses. Sensitivity analyses were conducted using log-transformed cIMT to improve the normality of the distribution, and the patterns of the associations were similar; therefore, results are not shown. In addition, nonparametric locally weighted scatterplot smoothing (LOESS) with a span of 1 (34) was used to validate and further describe the linear association between urinary MMA% and cIMT. We examined the assumption of a nonlinear effect of arsenic exposure and urinary arsenic metabolite indices by including higher-order polynomial terms for arsenic exposure and arsenic metabolic profile variables in the models, but we found no indication of any nonlinear relationship.

Coefficients were adjusted for cIMT-related risk factors, including sex, age, baseline body mass index (BMI; weight (kg)/height (m)²), educational attainment, and smoking status, that may influence the health effects of arsenic exposure in the population (27, 35, 36). In a separate model, we also included potential intermediate risk factors, such as systolic blood pressure and diabetes, which may mediate the development of atherosclerosis induced by arsenic exposure. Age, BMI, educational attainment, and systolic blood pressure were entered into the models as continuous variables. Total urinary arsenic level was expressed per gram of urinary creatinine to adjust for variation in hydration status (27, 37–43). We also used total urinary arsenic in µg/L with and without creatinine as a separate independent variable in the models (44, 45). For the associations of well-water arsenic and urinary arsenic with cIMT, we additionally adjusted for change in urinary arsenic between visits, which was associated with baseline well arsenic status and may be related to health effects. Continuous variables for change in urinary arsenic between baseline and the first follow-up and between the first and second follow-ups were included, and separate dummy variables were used for missing data at the first (n = 13) or second (n = 25) follow-up visit under a “missing at random” assumption. We explored the association between change in urinary arsenic level and cIMT. Additional adjustment for having switched wells since baseline did not change the effect estimates appreciably (data not shown). Sensitivity analyses were conducted using robust standard errors from generalized estimating equations models to account for some clustering of subjects using the same index well, and the results were nearly identical (data not shown). Analyses were also conducted after excluding subjects with high blood pressure (defined as systolic blood pressure of ≥140 mm Hg and/or diastolic blood pressure of ≥90 mm Hg (46)), diabetes, and a history of cardiovascular disease at baseline, as well as those diagnosed with coronary heart disease and stroke during follow-up.

We conducted stratified analyses to evaluate whether the associations of arsenic exposure and urinary arsenic metabolite indices with cIMT differed by sex, age, baseline BMI, and smoking status, since these factors may indicate susceptibility to health effects of arsenic (27, 35, 36). We tested for interaction between arsenic exposure and each of these potential effect modifiers by including a cross-product term with a dichotomized effect modifier and well-water arsenic or urinary arsenic expressed as a continuous variable in the models. The P value for the cross-product term was used to judge the statistical significance of interaction. We also conducted stratified analyses to assess whether the association between arsenic exposure and cIMT differed by arsenic methylation capacity, as indicated by levels of urinary InAs%, MMA%, DMA%, PMI, and SMI, with population median values used as the cutpoints. All analyses were performed using SPSS 19.0 software (SPSS Inc., Chicago, Illinois) and R, version 2.13.1 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

The proportion of participants with higher cIMT was greater among men than among women (Table 1). Average age, BMI, systolic blood pressure, and diastolic blood pressure at baseline were higher among persons in higher quintiles of cIMT. The proportion of ever smokers was greater among participants with higher cIMT (P < 0.001). Average MMA% in urine and PMI were higher, whereas average SMI was lower, in participants with higher cIMT. There was no evidence that well-water arsenic was associated with duration of well use prior to baseline, sex, age, baseline BMI, educational attainment, smoking status, systolic blood pressure, diastolic blood pressure, or diabetes status (Web Table 1). More details are available in the Web Appendix.

Table 1.

Demographic, Lifestyle, and Arsenic Exposure Characteristics of Participants by Quintile of Carotid Intima-Media Thickness, Health Effects of Arsenic Longitudinal Study, Bangladesh, 2000–2011^a

	Total (n = 959)				Quintile of cIMTb										P Valuec
					1 (n = 192)		2 (n = 201)		3 (n = 193)		4 (n = 182)		5 (n = 191)
	%	Mean	Median	10th–90th Percentiles	%	Mean	%	Mean	%	Mean	%	Mean	%	Mean
cIMT, µm		785.7	766.3	677.5–913.8		670.7		727.1		769.7		823.9		942.8
Baseline characteristics
Male sex	40.3				25.5		29.4		37.8		48.1		61.8		<0.001
Age, years		39.1	38	26–52		31.8		36.2		38.8		42.5		46.6	<0.001
Body mass indexd		20.0	19.3	16.3–24.7		19.3		20.1		20.3		20.2		20.4	0.034
Education, years		3.1	1	0–9		3.4		2.7		3.0		3.0		3.2	0.402
Ever smoking
Men	75.6				61.2		67.8		68.5		77.0		89.0		<0.001
Women	8.6				3.5		7.0		5.0		12.8		21.9		<0.001
Systolic blood pressure, mm Hg		116.9	115	97–137		111.1		113.6		116.3		118.7		124.9	<0.001
Diastolic blood pressure, mm Hg		75.2	74	62–88		73.0		73.7		75.1		76.5		78.0	<0.001
Diabetes	2.0				1.6		0.0		1.0		1.7		5.8		0.001
Well-water arsenic, µg/L		81.1	41	1–225		86.3		73.7		80.6		84.6		80.7	0.784
Duration of well use prior to  baseline, years		7.8	6	3–15		7.1		7.3		7.7		7.9		9.0	0.003
Urinary arsenic, µg/g creatinine		259.5	179.2	60.3–538.6		256.5		243.9		253.9		235.3		307.5	0.320
InAs%		15.5	14.5	8.8–23.8		16.6		15.7		14.9		15.4		14.6	0.074
MMA%		13.0	12.6	6.9–20.0		11.9		12.6		12.9		13.2		14.4	<0.001
DMA%		71.6	72.4	60.3–81.7		71.5		71.6		72.2		71.4		71.0	0.790
PMI		0.98	0.89	0.44–1.55		0.79		1.04		0.95		1.03		1.12	0.005
SMI		6.7	5.7	3.1–11.4		7.3		6.8		6.9		6.4		5.9	0.015
Follow-up characteristics
Time between baseline and cIMT  measurement, years		7.2	8.7	3.8–9.7		7.7		7.0		7.1		6.8		7.3	0.006
Urinary arsenic at first follow-up,  µg/g creatinine		210.5	146.9	58.0–449.0		218.8		198.6		220.9		194.0		220.2	0.519
Urinary arsenic at second follow-up,  µg/g creatininee		204.9	151	51–434		206.9		189.7		203.2		201.5		221.7	0.762
Change in urinary arsenic between  baseline and the first follow-up		−50.6	−15	−232.3–108.0		−38.1		−47.1		−34.7		−43.1		−90.1	0.493
Change in urinary arsenic between  the first and second follow-upse		−12.0	−4.0	−143.4–116.8		−17.3		−23.4		−40.5		−0.4		23.8	0.032
Age at cIMT measurement, years		46.3	45.5	35.2–58.8		39.5		43.2		45.9		49.3		53.9	<0.001

Abbreviations: cIMT, carotid intima-media thickness; DMA, dimethylarsinic acid; InAs, inorganic arsenic; MMA, monomethylarsonic acid; PMI, primary methylation index; SMI, secondary methylation index.

^a Data were missing on sex, baseline age, body mass index, smoking status, educational attainment, systolic blood pressure, diastolic blood pressure, and diabetes status for 1, 1, 3, 1, 1, 4, 5, and 4 subjects, respectively; on baseline well arsenic, urinary arsenic, and prebaseline duration of well use for 14, 1, and 35 subjects, respectively; and on age at cIMT measurement for 1 subject. Data were available on urinary InAs%, MMA%, DMA%, PMI, and SMI for only 782, 782, 782, 781, and 779 subjects, respectively, and on urinary arsenic at the first and second follow-ups for only 946 and 569 subjects, respectively.

^b Quintile 1: 558–703 µm; quintile 2: 704–748 µm; quintile 3: 749–793 µm; quintile 4: 794–860 µm; quintile 5: 861–1,243 µm.

^c P values were computed with the chi-square test or analysis of variance.

^d Weight (kg)/height (m)².

^e In the original cohort only.

In linear regression analysis, sex, age, baseline BMI, smoking status, and systolic blood pressure were significantly related to cIMT levels in this population (Web Table 2). There was a positive association between baseline urinary arsenic level and cIMT (Table 2). Every 1-SD increase in urinary arsenic concentration (357.9 µg/g creatinine) was related to an 11.7-µm (95% confidence interval (CI): 1.8, 21.6) increase in cIMT. The association between well arsenic and cIMT was similar but not statistically significant. In stratified analyses, although the positive association between baseline well arsenic and cIMT was stronger in men, in ever smokers, and in participants with a lower BMI, none of the interactions were statistically significant (all P's for interaction > 0.10). The pattern of the positive association between urinary arsenic and cIMT in stratified analyses was similar, although the significance of the association was not consistent in some subgroups, possibly because of the greater variation in urinary arsenic levels. In sum, despite some differences, the relationship between urinary arsenic and cIMT did not differ appreciably across groups. Sensitivity analyses using different adjustment methods for urinary creatinine generated similar results (Web Table 3). There was no apparent association between visit-to-visit change in urinary arsenic level during follow-up and cIMT. The associations remained after excluding subjects with high blood pressure, diabetes, and history of cardiovascular disease at baseline, as well as those diagnosed with coronary heart disease and stroke during follow-up. For instance, every 1-SD increase in baseline well arsenic and urinary arsenic was associated with a 6.8-µm (95% CI: 1.3, 12.3) and 14.5-µm (95% CI: 4.3, 24.6) difference in cIMT, respectively, after exclusion of subjects with any of the above-mentioned conditions.

Table 2.

Difference in Carotid Intima-Media Thickness in Relation to a 1-Standard-Deviation Increase in Arsenic Exposure Variables, Health Effects of Arsenic Longitudinal Study, Bangladesh, 2000–2011

	Baseline Well-Water Arsenic, µg/L				Baseline Urinary Arsenic, µg/g Creatinine
	No. of Subjects	βa	95% CI	P Value	No. of Subjects	βa	95% CI	P Value
Overall
Model 1b	943	4.0	−1.4, 9.3	0.148	956	11.0	0.8, 21.1	0.034
Model 2c	937	5.1	−0.2, 10.3	0.058	950	11.7	1.8, 21.6	0.020
Sex
Men
Model 1	380	10.1	−0.2, 20.3	0.055	386	9.9	−9.7, 29.4	0.321
Model 2	376	11.5	1.6, 21.3	0.023	382	13.3	−5.4, 32.0	0.164
Women
Model 1	563	2.1	−4.1, 8.3	0.502	570	9.6	−2.4, 21.5	0.116
Model 2	561	2.6	−3.4, 8.7	0.392	568	9.9	−1.8, 21.6	0.096
Body mass indexd
≤19.2
Model 1	473	10.0	2.8, 17.1	0.007	478	11.6	−1.0, 24.3	0.072
Model 2	468	10.4	3.3, 17.4	0.004	473	11.8	−0.6, 24.2	0.062
>19.2
Model 1	470	−1.7	−9.8, 6.4	0.676	478	7.3	−10.0, 24.5	0.409
Model 2	469	0.2	−7.7, 8.1	0.963	477	9.0	−7.7, 25.8	0.289
Age, years
≤45
Model 1	471	4.4	−2.8, 11.5	0.234	478	13.1	−0.5, 26.7	0.059
Model 2	468	5.3	−1.8, 12.4	0.143	475	15.1	1.7, 28.5	0.028
>45
Model 1	472	1.6	−7.7, 11.0	0.732	478	8.7	−8.5, 25.9	0.319
Model 2	469	3.5	−5.3, 12.4	0.432	475	8.9	−7.3, 25.2	0.281
Smoking status
Never smoker
Model 1	607	2.0	−3.9, 8.0	0.501	615	6.6	−5.2, 18.4	0.275
Model 2	605	2.6	−3.3, 8.4	0.391	613	6.5	−5.1, 18.1	0.273
Ever smoker
Model 1	336	10.8	−0.3, 21.9	0.057	341	14.5	−5.4, 34.4	0.152
Model 2	332	13.5	2.9, 24.2	0.013	337	18.9	−0.1, 37.9	0.051

Abbreviations: CI, confidence interval; cIMT, carotid intima-media thickness.

^a Difference in cIMT (µm) in relation to a 1-standard-deviation increase in well-water arsenic (102.0 µg/L) or urinary arsenic (357.9 µg/g creatinine) concentration.

^b Results were adjusted for all of the stratifying variables in the table, as well as age at cIMT measurement, baseline educational attainment, and change in urinary arsenic level between visits.

^c Results were additionally adjusted for systolic blood pressure and diabetes status.

^d Weight (kg)/height (m)².

We observed a positive relationship between urinary MMA% and cIMT (Table 3). For every 10% increase in MMA%, there was an increase of 12.1 µm (95% CI: 0.4, 23.8) in cIMT. The association between urinary MMA% and cIMT appeared to take the form of a monotonic dose-response, as shown by the LOESS analyses (Figure 1), and there was no evidence of deviation from a linear relationship. Urinary DMA% was inversely associated with cIMT, and every 10% increase in DMA% was related to a decrease of 6.3 µm (95% CI: −12.8, 0.2) in cIMT. There was no apparent association of urinary InAs%, PMI, or SMI with cIMT. The patterns of the associations did not differ by sex, age, baseline BMI, or smoking status. For instance, every 10% increase in MMA% was related to cIMT increases of 14.2 µm (95% CI: −6.1, 34.6) and 8.3 µm (95% CI: −5.3, 21.8) in men and women, respectively (data not shown).

Table 3.

Difference in Carotid Intima-Media Thickness in Relation to Urinary Arsenic Metabolite Indices, Health Effects of Arsenic Longitudinal Study, Bangladesh, 2000–2011

Variable	No. of Subjectsa	βb	95% CI	P Value
InAs%	774	4.1	−4.1, 12.3	0.326
MMA%	774	12.1	0.4, 23.8	0.042
DMA%	774	−6.3	−12.8, 0.2	0.057
PMI	773	1.5	−5.2, 8.1	0.664
SMI	771	−1.2	−2.8, 0.4	0.140

Abbreviations: CI, confidence interval; cIMT, carotid intima-media thickness; DMA, dimethylarsinic acid; InAs, inorganic arsenic; MMA, monomethylarsonic acid; PMI, primary methylation index; SMI, secondary methylation index.

^a Data were missing on PMI for 1 subject (because InAs% was 0) and on SMI for 3 subjects (because MMA% was 0).

^b Difference in cIMT (µm) in relation to a 10% increase in InAs%, MMA%, or DMA% or a 1-unit increase in PMI or SMI, after adjustment for sex, age at cIMT measurement, baseline body mass index, smoking status (never, past, or current smoker), educational attainment, systolic blood pressure, and diabetes status.

Figure 1.

Relationship of urinary monomethylarsonic acid percentage (MMA%) to carotid intima-media thickness (cIMT), as estimated using locally weighted scatterplot smoothing (LOESS), Health Effects of Arsenic Longitudinal Study, Bangladesh, 2000–2011. The solid black line represents the LOESS smoothing fit of cIMT level on MMA%; the dashed lines represent the 95% pointwise confidence intervals for the smoothing cIMT levels; and the solid gray line represents the univariate linear regression fit of cIMT level on MMA%.

Among participants with higher urinary InAs%, MMA%, and PMI or lower SMI, the associations between well-water arsenic and cIMT were stronger (Figure 2). The increases in cIMT in relation to a 1-SD increase in well-water arsenic were 7.5 µm (95% CI: 0.2, 14.9) in persons with higher InAs%, 12.5 µm (95% CI: 3.3, 21.8) in persons with higher MMA%, 9.1 µm (95% CI: 0.7, 17.5) in persons with higher PMI, and 11.9 µm (95% CI: 2.7, 21.0) in persons with lower SMI.

Figure 2.

Association between well-water arsenic level and carotid intima-media thickness (μm) according to urinary arsenic metabolite indices, Health Effects of Arsenic Longitudinal Study, Bangladesh, 2000–2011. The β coefficient was adjusted for sex, age (years), baseline body mass index, smoking status (never, past, or current smoker), educational attainment, systolic blood pressure, diabetes status, and change in urinary arsenic concentration (μg/g of creatinine) between visits. CI, confidence interval; DMA, dimethylarsinic acid; InAs, inorganic arsenic; MMA, monomethylarsonic acid; PMI, primary methylation index; SMI, secondary methylation index.

DISCUSSION

We found a positive association between past arsenic exposure assessed at baseline and cIMT assessed during follow-up. There was a dose-response relationship between the proportion of MMA in baseline urine and cIMT measured on average 7.2 years later. In addition, the association between baseline well arsenic and cIMT was more significant in subjects with higher urinary InAs%, MMA%, and PMI or lower SMI.

Several studies from Taiwan previously found a positive association between water arsenic and carotid atherosclerosis (18–24) among participants who had been exposed to arsenic 20–30 years earlier. In the present study, the association between arsenic exposure and cIMT was stronger when baseline total urinary arsenic was used as an exposure variable. Total urinary arsenic, which correlated well with well-water arsenic (r = 0.72), has been used as a long-term biomarker for arsenic exposure from drinking water in the cohort (26) and may better capture exposure variation in the cohort, reflecting internal dose. To our knowledge, the present study is the first to have found a positive association between urinary arsenic and cIMT. The positive association between baseline urinary arsenic and cIMT measured 7 years later indicates a long-term effect of arsenic exposure on cIMT.

A growing number of mechanistic studies have shown that trivalent MMA (MMA^III) is more toxic than InAs or any of the pentavalent metabolites (47, 48). In a recent study from Taiwan with 304 subjects, Huang et al. (49) reported that cases with carotid atherosclerosis had an insignificantly greater urinary MMA% than controls. In the present study, we found a linear dose-response relationship between urinary MMA% and cIMT, suggesting an effect of incomplete or suboptimal arsenic methylation capacity on atherosclerosis. We also observed that the associations between well arsenic and cIMT in persons with higher MMA%, higher PMI, or lower SMI were stronger, indicating that the effect of arsenic exposure from drinking water on atherosclerosis risk is enhanced in persons with suboptimal or incomplete arsenic methylation.

It has been reported that a 100-µm increase in IMT at the low end of IMT could relate to a 50% increased risk of coronary heart disease (50). Based on this finding, the cIMT difference (12-µm increase in cIMT) associated with a 1-SD increase in urinary arsenic level (approximately equivalent to a 100-µg/L well arsenic concentration) or a 10% increase in urinary MMA% could translate to a 5% increased risk of coronary heart disease, although more research is needed for confirmation. Given that coronary heart disease is prevalent, a small increase in the risk of a preclinical marker could translate to a substantial number of cases.

The exact mechanisms by which arsenic leads to atherosclerosis are not fully understood. In human umbilical vein endothelial cells, arsenic was found to enhance the effect of tumor necrosis factor α in increasing vascular cell adhesion molecule 1 protein expression and to subsequently induce vascular inflammation and vascular disease (51). In apolipoprotein E-knockout mice, a well-established animal model for arsenic-induced atherosclerosis, exposure to a moderate concentration of arsenic increased formation of less stable and more dangerous atherosclerotic lesions by inhibiting genes that regulate lipid homeostasis and by increasing the circulating levels of several proinflammatory cytokines (52). Li et al. (18) found that carotid atherosclerosis after arsenic exposure was associated with low serum paraoxonase activity. A study in Taiwan demonstrated the joint effects of apolipoprotein E with monocyte chemoattractant protein 1 on the low-density lipoprotein cholesterol inflammatory reaction in atherosclerosis progression after arsenic exposure (22). In cross-sectional studies of the baseline data, we observed positive associations between arsenic exposure and plasma levels of soluble cell adhesion molecules, markers of endothelial dysfunction for risk of cardiovascular disease (37, 53). Arsenic exposure has also been associated with prolonged Q-T interval, a marker of subclinical atherosclerosis, in studies conducted in the United States and Inner Mongolia (People's Republic of China) (7, 54) and more recently in our study population (8).

There were several limitations of this study. First, although participants had on average used the baseline wells for 7.8 years, we did not have complete histories of lifetime exposure, and some participants did not use the baseline well exclusively. The potential misclassification of exposure, though it should not have been differential by cIMT status, may have contributed to the weaker association seen between well-water arsenic and cIMT. Second, urinary arsenic metabolites were measured in 1 spot urine sample, which may not be reflective of the individual's long-term usual level. However, the literature suggests that an individual's arsenic methylation efficiency is stable over time (55). Third, the study results might not be generalizable to other populations with a different profile of risk factors for cardiovascular disease. The study population consisted of mostly lean people, with a mean BMI of 20. We observed that the positive associations between arsenic exposure and cIMT were more pronounced in participants with a lower BMI. Lower BMI reflects poorer nutritional status in rural Bangladesh, which may be associated with lower intake of nutrients involved in the etiology of vascular disease and arsenic metabolism and detoxification (56, 57). We did not aim to assess the role of specific nutritional intake in the present study. Future studies are needed to address whether intakes of certain nutrients, especially those relevant to arsenic methylation capacity, may contribute to the stronger association in subjects with lower BMI.

Importantly, cIMT measurements were not conducted for all of the selected persons. However, it is unlikely that the subjects with high levels of both baseline arsenic exposure and cIMT were preferentially included (a necessary condition for the presence of bias). The distributions of demographic, lifestyle, and arsenic exposure variables in the study population and in the overall cohort were very similar (data not shown). There is no evidence that arsenic exposure is related to longer or better survival from cardiovascular disease (incidence-prevalence bias), and the associations remained after we excluded cardiovascular disease cases and persons with diabetes. Atherosclerosis is a chronic disease which is likely to develop over a long period of time as a result of past and persistent exposure. However, the absence of cIMT measurement at recruitment makes it impossible to firmly establish a temporal relationship between arsenic exposure and cIMT. We did not collect information on lipid profiles at baseline. However, the available literature does not suggest a positive association between arsenic exposure and lipid profiles or between lipid profiles and cardiovascular disease related to arsenic exposure (58, 59).

In summary, we observed a positive association of arsenic exposure and urinary MMA% with cIMT. The association between well-water arsenic and cIMT was modifiable by incomplete methylation efficiency, indicated by higher urinary MMA%, a higher ratio of urinary MMA to InAs, or a lower ratio of urinary DMA to MMA. These findings could be of public health importance, as they help to identify arsenic exposure from drinking water as an environmental risk factor for cardiovascular disease and raise the possibility that atherosclerosis in an arsenic-exposed population could be reduced by enhancing MMA methylation.