Arsenic Exposure and Hypertension: A Systematic Review

1Hypertension is a major risk factor for mortality and morbidity worldwide. Risk factors for hypertension include high salt intake, increased body mass index (BMI), genetic predisposition, and exposure to psychosocial stress. Additional evidence, however, suggests that environmental factors play a role in hypertension development. The identification and mitigation of environmental exposures related to hypertension could contribute to reducing the worldwide burden of hypertension-related disease.

2Among environmental exposures, epidemiologic and experimental evidence supports the possibility that arsenic plays a role in hypertension and other cardiometabolic diseases. Arsenic-contaminated drinking water represents a major public health problem internationally. The World Health Organization and U.S. Environmental Protection Agency (EPA) standard for arsenic levels in drinking water is 10 μg/L. In the United States alone, millions of persons are exposed to arsenic concentrations > 10 μg/L; whereas persons in Bangladesh, China, India, Cambodia, Ghana, Argentina, Mexico, and other countries around the world are exposed to arsenic levels in drinking water that are well beyond 10 μg/L. Epidemiologic studies conducted in arsenic-endemic areas in Taiwan and Bangladesh have found a positive relationship between inorganic arsenic exposure from drinking water and hypertension. Experimental studies have indicated that arsenic exposure may be involved in the development of hypertension through the promotion of inflammation, oxidative stress, and endothelial dysfunction.

3To evaluate the potential relationship between arsenic and hypertension, we conducted a systematic review of epidemiologic studies that have investigated the association between inorganic arsenic exposure (using environmental measures or biomarkers) and hypertension outcomes [using hypertension status and systolic and diastolic blood pressure (SBP and DBP, respectively)].

4Search strategy and data abstraction. We searched PubMed (http://www.ncbi.nlm.nih.gov/pubmed/, Embase (http://www.embase.com/home), and TOXLINE (http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen?TOXLINE) databases to find all published observational studies evaluating the relationship between arsenic exposure with hypertension or BP levels using the free text and Medical Subject Headings (MeSH) terms “arsenic,” “arsenicals,” “arsenate,” or “arsenite” and “hypertension” or “blood pressure.” The search period was January 1966 through March 2011 with no language restrictions (Figure 1).

5Two investigators (L.N.A. and M.R.J.) reviewed each paper and applied the study selection criteria (Figure 1). Epidemiologic studies with data on arsenic exposure and hypertension outcomes were included. We excluded nonoriginal reports, experimental studies, case reports and case series, and studies without measures of arsenic exposure or hypertension end points. We also excluded one study that used hypertension mortality as the only end point and two reports that used the same study population as another included study. The two investigators, L.N.A. and M.R.J., independently abstracted the study data, including design, study population (location, age, and sex distribution), sample size, arsenic assessment and exposure levels, hypertension outcomes, study results (measures of association), and potential confounders accounted for in the statistical analysis. Authors were contacted for information unavailable in the published reports. For studies with multiple levels of adjustment, we abstracted the measure of association obtained from the model adjusted for the most covariates after confirmation that adjustment did not markedly modify the conclusions of any individual study. For studies that were not in English (138 of 865), the full text of the article was translated by a native speaker if the information in the abstract was insufficient to include/exclude the article. Discrepancies were resolved by consensus. The quality of the included studies was evaluated by adapting the criteria developed by and.

6Statistical analysis. For studies that reported hypertension, we abstracted or derived odds ratios (ORs) and prevalence ratios for hypertension and their standard errors from the published data. For three studies with hypertension data but no available measures of association, we estimated the OR and 95% confidence interval (CI) for hypertension by arsenic categories using the number of cases and noncases in the exposed and unexposed groups. For summary purposes, we pooled OR estimates comparing hypertension in the highest and lowest categories of arsenic exposure from individual studies using an inverse-variance weighted random-effects model. Pooled ORs were calculated for all studies and separately for studies conducted in populations exposed to moderate-to-high arsenic levels and for studies conducted in populations exposed to low arsenic levels. Heterogeneity was quantified with the I2 statistic, an index that describes the proportion of the total variation in pooled estimates due to heterogeneity. The relative influence of each study on pooled estimates was estimated by omitting one study at a time. Finally, we assessed publication bias using funnel plots. For studies that reported hypertension results for three or more arsenic categories, we evaluated the dose–response relationship over the range of arsenic levels. All statistical analyses were performed using Stata software, version 11.0 (StataCorp, College Station, TX, USA).

7For studies that reported SBP and DBP levels, we abstracted or derived the difference in BP levels comparing the highest and lowest categories of arsenic exposure. Because the number of studies was small and because the largest study did not provide enough information to calculate CIs, these results are presented descriptively, and no pooled estimate was calculated.

8Study characteristics. Eleven studies, published between 1995 and 2011, were identified (Table 1). All studies meeting the inclusion criteria were cross-sectional and published in English. Combined, the studies covered arsenic exposure and hypertension outcomes for > 20,000 individuals. Eight studies were conducted at moderate to high levels of exposure (average levels in drinking water ≥ 50 μg/L or occupational studies), and three studies were conducted at low levels of exposure (average levels in drinking water < 50 μg/L). Ten studies were conducted in general populations (two from Taiwan, two from Bangladesh, two from Inner Mongolia, two from the United States, one from Turkey, and one from Iran). One study was conducted in an occupational setting in Denmark. Five studies measured arsenic concentrations in drinking water, three compared areas of high and low arsenic concentrations in drinking water, two studies used biomarkers (hair,; urine, ), and one study assigned arsenic exposure based on job title. Eight studies assessed hypertension as the end point of interest, five studies reported differences in mean SBP, and four studies reported differences in mean DBP.

9Quality assessment. Five studies measured arsenic in drinking water at the individual level; three of these studies measured individual arsenic exposure based on measured well water concentrations, and two studies used a biomarker of exposure (Table 2). Five studies defined hypertension based on established cutoffs for SBP and DBP levels measured with a standardized protocol and self-reported physician diagnosis or antihypertensive treatment. Five of the 11 studies did not adjust for potential confounders. Other studies adjusted at least for age, sex, and BMI.

10ORs estimates for hypertension. For the association of hypertension with arsenic exposure, five of the eight studies found a positive association. Among the studies that assessed hypertension at moderate to high levels of exposure, the OR estimates comparing highest with lowest arsenic exposure groups ranged from 0.71 (95% CI: 0.18, 2.63) in a small study in Turkey to 16.5 (95% CI: 2.8, 668.5) in a study in Inner Mongolia (Figure 2). The two studies from Bangladesh provided inconsistent results: an OR of 3.0 (95% CI: 1.5, 5.8) in the study by and an OR of 1.02 (95% CI: 0.84, 1.23) in the study by Chen Y et al. (2007). Among the studies that assessed hypertension at low levels of exposure, the OR estimates comparing highest with lowest arsenic exposure groups ranged from 1.17 (95% CI: 0.75, 1.83) in a study in the general U.S. population to 2.00 (95% CI: 1.21, 3.31) in a study in central Taiwan.

11The pooled OR of hypertension comparing the highest and lowest arsenic exposure categories in the eight studies with available information on hypertension was 1.27 (95% CI: 1.09, 1.47; p-value for heterogeneity = 0.001; I2 = 70.2%). The corresponding pooled OR in the five studies with moderate to high arsenic exposure was 1.15 (95% CI: 0.96, 1.37; p-value for heterogeneity = 0.002; I2 = 76.6%), with the study by Chen Y et al. (2007) being highly influential. Excluding that study, the pooled OR was 2.57 (95% CI: 1.56, 4.24; p-value for heterogeneity = 0.13; I2 = 46.6%). The pooled OR comparing the highest and lowest arsenic exposure categories in the three studies with low arsenic exposure was 1.56 (95% CI: 1.21, 2.01; p-value for heterogeneity = 0.27; I2 = 24.6%). We also restricted the overall pooled analysis to studies with multivariable adjusted ORs (pooled OR = 1.22; 95% CI: 1.04, 1.42), studies with a standard hypertension definition (pooled OR = 1.21; 95% CI: 1.03, 1.42), and studies with individual assessment of arsenic exposure (pooled OR = 1.19; 95% CI: 1.02, 1.38). Funnel plots did not suggest the presence of publication or related biases (data not shown).

12We evaluated the dose response for six studies with ORs reported for three or more categories (Figure 3). Among them, the Chen Y et al. (2007) study in Bangladesh showed no dose–response relationship. Compared with the baseline category, the other study from Bangladesh and the study from Taiwan showed increased prevalence of hypertension for most of the arsenic exposure categories. Studies conducted at low levels of exposure in drinking water showed an increased prevalence of hypertension throughout the range of arsenic exposure levels, although the association was not statistically significant for the intermediate arsenic categories.

13Difference in BP level estimates. For the association of arsenic exposure with BP levels, three of five studies found a positive association with SBP, and two of four studies found a positive association with DBP (Figure 4). The difference in BP levels comparing the highest and lowest arsenic exposure categories ranged from –0.79 to 30.0 mmHg for SBP and from –0.65 to 11.04 mmHg for DBP. Only two studies adjusted for hypertension risk factors.

14This systematic review identified an association between arsenic exposure and the prevalence of hypertension. The association was present both in studies conducted in areas with moderate-to-high arsenic exposure levels and in studies conducted in areas with low exposure levels. A clear dose–response was observed in several studies, and experimental evidence supports the hypertensive effects of arsenic. The interpretation of this association regarding the causal effect of arsenic on hypertension, however, is limited by the small number of studies, the heterogeneity across studies, and the absence of prospective evidence. In addition, some studies were affected by additional methodological limitations such as the lack of standard hypertension definitions, individual assessment of arsenic exposure, or appropriate adjustment for relevant confounders. The evidence is particularly scarce for low levels of exposure and for evaluating the association with SBP and DBP levels as continuous outcomes. Overall, the evidence is suggestive but insufficient to infer a causal relationship between environmental arsenic exposure and hypertension.

15Two studies from areas with high arsenic levels in drinking water in southwestern Taiwan and Bangladesh and two studies conducted in areas with low levels of arsenic in drinking water in Wisconsin and central Taiwan showed consistent associations of arsenic exposure with the prevalence of hypertension. These four studies also showed a consistent dose–response increase in the prevalence of hypertension with increasing arsenic exposure.

16Discrepancies in the association between arsenic and the prevalence of hypertension were observed in four studies. The study with the strongest association (OR = 16.54; ) and the study with the inverse association (OR = 0.71; ) had small numbers of cases, provided no definition of hypertension, and incorporated no adjustment for relevant confounders. Both studies were highly imprecise with large CIs. The two null studies were large high-quality studies conducted in Bangladesh and the United States. The study in Bangladesh found no dose–response relationship, despite assessing arsenic at the individual level and defining hypertension based on BP measures. However, this study did find an association between arsenic levels in drinking water with systolic hypertension and pulse pressure levels among participants with low folate and vitamin B intake levels, whereas subgroup analyses by folate and vitamin B concentrations were conducted in the study in the general U.S. population, with no differences. In the study conducted among the general U.S. population, the association between arsenic exposure and hypertension was not statistically significant, and it was consistent with no association. However, the magnitude of the association was compatible with a small increased prevalence of hypertension and consistent with the dose–response trend observed in other studies conducted at low-to-moderate exposure levels in Wisconsin and central Taiwan.

17The potential association between exposure to inorganic arsenic and the development of hypertension is supported by experimental and mechanistic evidence, especially at high exposure levels. Arsenic promotes inflammation activity, oxidative stress, and endothelial dysfunction through several mechanisms including the activation of stress response transcription factors such as activator protein-1 and nuclear factor-κB. In vitro, arsenite altered vascular tone in blood vessels by suppressing vasorelaxation and increased the expression of cyclooxygenase-2 in endothelial cells. In animal models, arsenite increased superoxide accumulation and impaired nitric oxide formation in endothelial cells. Finally, the hypertensive effects of arsenic could be related to the possible chronic kidney effects of arsenic. Additional experimental studies using arsenic exposure levels relevant to human populations are needed to characterize the etiopathogenesis of potential hypertensive effects of arsenic.

18This is the first systematic review and meta-analysis evaluating the relationship between arsenic exposure and hypertension end points. We identified a positive association between elevated arsenic exposure and the prevalence of hypertension, but the implications of this association from a causal perspective are unclear because of the limited number of studies as well as the studies’ cross-sectional design, and methodological limitations. Prospective cohort studies in populations exposed to a wide range of arsenic exposure levels, from low through moderate-to-high levels of exposure, are needed to better characterize the relationship between arsenic and hypertension. Because of the widespread exposure to arsenic worldwide and the high burden of disease caused by hypertension, it is important that high-quality prospective studies are conducted with individual level assessment of arsenic exposure and standardized measurements of BP. The studies should evaluate the shape of the dose response and whether the magnitude of the association is different in susceptible populations, including populations with nutritional deficiencies. If the hypertensive effects of arsenic are confirmed, they could partly explain the association between arsenic and cardiovascular disease. Given the widespread arsenic exposure through drinking water and food, even a modest effect of arsenic on hypertension could have a substantial impact on morbidity and mortality.