Arsenic exposure, inflammation, and renal function in Bangladeshi adults: effect modification by plasma glutathione redox potential

Abstract

Exposure to arsenic (As) in drinking water is a widespread public health problem leading to increased risk for multiple outcomes such as cancer, cardiovascular disease, and possibly renal disease; potential mechanisms include inflammation and oxidative stress. We tested the hypothesis that As exposure is associated with increased inflammation and decreased estimated glomerular filtration rate (eGFR) and examined whether the effects of As were modified by plasma glutathione (GSH), glutathione disulfide (GSSG), or the reduction potential of the GSSG/2GSH pair (E_hGSH). In a cross-sectional study of N = 374 Bangladeshi adults having a wide range of As exposure, we measured markers of inflammation (plasma C-reactive protein (CRP), α-1 acid glycoprotein (AGP)), renal function (eGFR), GSH, and GSSG. In covariate-adjusted models, a 10% increase in water As, urinary As adjusted for specific gravity (uAs), or blood As (bAs) was associated with a 0.74% (p = 0.01), 0.90% (p = 0.16), and 1.39% (p = 0.07) increase in CRP, respectively; there was no association with AGP. A 10% increase in uAs or bAs was associated with an average reduction in eGFR of 0.16 (p = 0.12) and 0.21 ml/min/1.73 m² (p = 0.08), respectively. In stratified analyses, the effect of As exposure on CRP was observed only in participants having E_hGSH > median (uAs p_Wald = 0.03; bAs p_Wald = 0.05). This was primarily driven by stronger effects of As exposure on CRP in participants with lower plasma GSH. The effects of As exposure on eGFR were not modified significantly by E_hGSH, GSH, or GSSG. These data suggest that participants having lower plasma GSH and a more oxidized plasma E_hGSH are at increased risk for As-induced inflammation. Future studies should evaluate whether antioxidant treatment lowers plasma E_hGSH and reduces risk for As-induced diseases.

Globally, the World Health Organization estimates that more than 200 million people are chronically exposed to arsenic (As)-contaminated drinking water above the safety standard of 10 μg/L [1]. Arsenic exposure is associated with increased risk for skin, lung, and bladder cancer, as well as noncancer outcomes including cardiovascular disease (CVD)¹, respiratory illness, and neurologic deficits [2]. Chronic kidney disease (CKD) is also emerging as a potential As-induced disease outcome [3]. We have observed in a cross-sectional study in Bangladesh that total urinary As (uAs), but not water As (wAs), was marginally associated with decreased estimated glomerular filtration rate (eGFR) [4].

Arsenic exposure has been shown to cause inflammation; this may be a mechanism for As-induced diseases, as inflammation is involved in the pathogenesis of many chronic diseases, including CVD, metabolic syndrome [5], CKD [6], and cancer [7]. Chronic exposure of mice to 100 ppb sodium arsenite in drinking water resulted in an increase in the acute-phase protein C-reactive protein (CRP) in the liver and kidney [8]. Increases in proinflammatory cytokines have also been observed in serum of arsenic-exposed animals [9,10]. Additionally, As exposure in humans has been associated with increased serum proinflammatory cytokines interleukin-6 (IL-6) and IL-8 [11] and CRP [12].

Numerous in vitro and in vivo studies have demonstrated that As toxicity is mediated by induction of oxidative stress [13]. Reactive oxygen species (ROS) that have increased in response to As include the superoxide anion radical and hydrogen peroxide [14]. Oxidative stress and inflammation are connected in a complex feedback cycle in which ROS can activate transcription factors that upregulate the expression of proinflammatory cytokines and both pro-oxidant and antioxidant enzymes [15]. Additionally, phagocytic leukocytes recruited to sites of inflammation express enzymes that contribute to oxidative damage, which can further amplify the inflammation [16]. Oxidative stress and inflammation are linked as causal agents in a variety of chronic diseases including cancer, CVD, and diabetes [17,18].

There is evidence to suggest that oxidative stress may mediate both As-induced inflammation and As-induced renal dysfunction. In certain mammalian cell models, arsenite exposure induces oxidative stress (i.e., increased ROS) and activates the proinflammatory transcription factor nuclear factor-κB (NF-κB), which can be prevented by administration of the antioxidant N-acetylcysteine (NAC) [19–21]. Arsenic exposure in rodent studies has resulted in increased ROS, lipid peroxidation, and protein carbonylation; decreased reduced glutathione (GSH) and increased glutathione disulfide (GSSG); and reduced antioxidant enzyme activity in kidney tissue [22–26]. The above-mentioned studies have also demonstrated that treatment of As-exposed rodents with antioxidants can either mitigate or prevent the oxidative and nephrotoxic effects observed with As administration alone. Collectively, this evidence suggests that antioxidants may protect against the inflammatory and nephrotoxic effects of As.

The Folate and Oxidative Stress (FOX) study [27] is a data-rich cross-sectional study of 378 Bangladeshi adults that was initially designed to assess the dose–response relationship between As exposure and oxidative stress. In this study, we previously found As exposure to be negatively associated with blood GSH concentrations, but not with blood GSSG [27], the plasma GSH/GSSG ratio [28], urinary 8-oxoG, or plasma protein carbonyls [29]. Plasma GSH and GSSG may correlate with tissue GSH and GSSG concentrations, and the reduction potential of the GSSG/2GSH couple (E_hGSH) may reflect the balance of pro-oxidants and antioxidants within tissues [30]. In the current analyses, we examine markers of inflammation and renal function in these study participants to test the hypotheses that As exposure is associated with inflammation and that the effects of As exposure on inflammation and renal function are greater among people with a more oxidized plasma E_hGSH. We also wished to examine whether inflammation mediates the effect of As on renal function, as inflammation is involved in the deterioration of renal function [6].

1. Materials and methods

1.1. Study participants

The FOX study was a cross-sectional study designed to examine the dose–response relationship between As exposure and oxidative stress, as described previously [27]. Briefly, N = 378 adults living in Araihazar, Bangladesh, were recruited in 2008 for this study. Individuals had to be between the ages of 30 and 65; not pregnant; not taking nutritional supplements; free of known diabetes, cardiovascular or renal disease, chronic obstructive pulmonary disease, or cancer; and drinking from their current well for at least 3 months. Participants were selected based on the As concentration of their wells, to include a wide range of As exposures (Group A, 0–10 μg/L (N = 76); Group B, 10–100 μg/L (N = 104); Group C, 100–200 μg/L (N = 86); Group D, 200–300 μg/L (N = 67); Group E, >300 μg/L (N = 45)). Participants missing information on wAs, BMI, specific gravity, plasma E_hGSH, or α-1 acid glycoprotein (AGP) were excluded from the current analysis, leaving a final sample size of N = 374.

Oral informed consent was obtained by our Bangladeshi field staff physicians, who read an approved assent form to the study participants. This study was approved by the Bangladesh Medical Research Council and the institutional review board of Columbia University Medical Center.

1.2. Collection of biospecimens

Spot urine samples were collected into 50-ml acid-washed polypropylene tubes and frozen at −20 °C. After initial processing of blood samples in the field clinic, blood and plasma aliquots were immediately frozen at −80 °C. Samples were transported to Dhaka on dry ice and again stored at −80 °C (blood and plasma) or −20 °C (urine). Samples were then packed on dry ice and flown to Columbia University.

1.3. Well-water As

Water samples were analyzed by high-resolution inductively coupled plasma mass spectrometry [31]. Water samples were collected in 20-ml polyethylene scintillation vials and acidified to 1% with high-purity Optima HCl (Fisher Scientific, Pittsburg, PA, USA) at least 48 h before analysis. A standard with an As concentration of 51 μg/L was run multiple times in each batch. The intra- and interassay coefficients of variation (CVs) for this standard were 6.01 and 3.76%, respectively.

1.4. Total uAs and specific gravity

The arsenic metabolites (arsenite (A^III), arsenate (As^V), monomethylarsonous acid plus monomethylarsonic acid (MMA^III+V), and dimethylarsinous acid plus dimethylarsinic acid (DMA^III+V)) were measured in urine by coupling HPLC to dynamic reaction cell inductively coupled plasma mass spectrometry, as described previously [32]. Total uAs was calculated as the sum of As^III, As^V, MMA^III+V, and DMA^III+V. The intra-assay CVs for urinary As^III, As^V, MMA, and DMA were 3.6, 4.5, 1.5, and 0.6%, respectively; the interassay CVs were 9.7, 10.6, 3.5, and 2.8%, respectively. We measured specific gravity by refractometer. Urinary As was adjusted for specific gravity (SG) using the following formula: uAs × (overall mean SG − 1)/(measured SG − 1). We refer to this adjusted variable as uAs-SG.

1.5. Total blood As (bAs)

As described previously [33], bAs concentrations were measured using a PerkinElmer Elan DRC II ICP-MS equipped with an AS 93+ autosampler. The intra- and interassay CVs were 2.1 and 4.9%, respectively.

1.6. Plasma glutathione and glutathione disulfide

Whole blood and plasma GSH and GSSG were measured by HPLC with fluorescence detection [34], as previously described [27]. Intra-assay CVs ranged from 5 to 10% and interassay CVs ranged from 11 to 18%.

1.7. Calculation of the redox potential

The E_hGSH was calculated using the Nernst equation, E_h = E_o + RT/nF ln[disulfide]/[thiol]². In this equation, E_o is the standard potential for the redox couple (−264 mV), R is the gas constant, T is body temperature in Kelvin, n is 2 for the number of electrons transferred, and F is Faraday’s constant [30]. The E_h represents the two-electron half-cell reduction potential of the GSSG/2GSH couple, and as such a more positive E_h value reflects a more oxidized redox state.

1.8. Plasma nutrients

Plasma folate and B12 were measured by radio-protein binding assay (SimulTRAC-S, MP Biomedicals) as previously described [27]. The intra- and interassay CVs for folate were 9 and 14%, respectively, and those for B12 were 5 and 9%, respectively. Plasma total homocysteine (tHcys) was measured by HPLC with fluorescence detection [35]. The intra- and interassay CVs for tHcys were 2 and 9%, respectively.

1.9. C-reactive protein and α-1 acid glycoprotein

CRP and AGP were assayed by ELISA with Quantikine ELISA kits from R&D Systems (Minneapolis, MN, USA). The intra- and inter-assay CVs were 4 and 3% for CRP, respectively, and 5 and 6% for AGP, respectively.

1.10. Plasma cystatin C and eGFR

Cystatin C was measured by ELISA according to the manufacturer’s protocol (R&D Systems Human Cystatin C Duoset, Catalog No. DY1196). We used a six-point standard curve with a high standard of 3000 pg/ml. Samples were diluted 1:2000 in phosphate-buffered saline with 10% fetal bovine serum (Sigma–Aldrich F6178). Recovery of the IFCC-certified reference material for serum cystatin C (ERM-DA 471/IFCC) was 104%. The intra- and interassay CVs were 7 and 10%, respectively. We calculated eGFR using the 2012 CKD-EPI Cystatin C equation [36].

1.11. Statistical analysis

We calculated descriptive statistics (means ± standard deviations for continuous variables and frequencies for categorical variables) for the total study population and stratified by the median of plasma E_hGSH. To reduce variation in plasma GSH and GSSG, we adjusted these variables for plasma GSH laboratory batch as a categorical variable using the residual method; these batch-adjusted variables were used in the calculation of plasma E_hGSH. Differences in characteristics between the E_hGSH strata were tested using the Wilcoxon rank-sum test for continuous variables and the χ² test for categorical variables. We used Spearman correlation coefficients to explore bivariate associations between CRP, AGP, eGFR, GSH, GSSG, and E_hGSH, adjusting for age, sex, and BMI, which are strongly associated with eGFR and the inflammatory biomarkers. We also used Spearman correlations to examine associations between the measures of As exposure (wAs, uAs-SG, and bAs) and GSH, GSSG, and E_hGSH, adjusting for covariates that were associated with at least one measure of As exposure (p < 0.1) and at least one glutathione variable (p < 0.1) in bivariate correlations.

For our primary analysis, we used linear regression models to examine the effect of the As exposure variables, wAs, uAs-SG, and bAs, on the outcomes CRP, AGP, and eGFR, controlling for potential confounders. We a priori decided to adjust for age, sex, and BMI in these models. Other covariates considered for inclusion were plasma folate, plasma B12, smoking, betel nut chewing, television ownership, education, systolic blood pressure, and diastolic blood pressure. Covariates were considered based on their Spearman correlations with the As exposure variables and the outcome variables, adjusted for age and sex. The control variables in the final models were those that were associated with CRP, AGP, or eGFR and with any one As exposure variable in bivariate correlations (p < 0.1), and resulted in an appreciable (>10%) change in the regression coefficient for the association between a predictor and an outcome. Variables with skewed distributions (wAs, bAs, uAs-SG, age, BMI, CRP, and AGP) were natural log transformed for linear regression analyses. To aid interpretation, for the models in which both the predictor of As exposure and the outcome (CRP or AGP) were log-transformed, we estimated the percentage change in the geometric mean of the outcome for a 10% increase in the As exposure variable. For the outcome of eGFR in its original scale, with log-transformed As variables as predictors, we estimated the mean change in eGFR for a 10% increase in the As exposure.

Effect modification by plasma E_hGSH was examined by repeating the regression analyses in strata of plasma E_hGSH (above and below the median). We used the Wald test to detect differences in the covariate-adjusted associations of As exposure with the outcome variables between plasma E_hGSH strata. To explore whether the effect modification was dependent on the cut-point for E_hGSH, we conducted a sensitivity analysis in which we repeated the linear regression analysis above and below every unique E_hGSH cut-point, starting at the 20th percentile of E_hGSH and ending at the 80th percentile, for a total of 224 cut-points. In the same manner we also checked effect modification by GSH and GSSG plasma concentrations.

All p values were two-sided. Analyses were performed using SAS 9.1 (SAS Institute, Cary, NC, USA) and R version 3.0.2.

2. Results

Characteristics of the study sample are reported in Table 1 for the entire study sample and separately by plasma GSH redox potential (E_hGSH) strata, which were derived by dividing plasma E_hGSH at the median. On average, participants having a more oxidized plasma E_hGSH (> −99.14) were older (p = 0.05), were more frequent betel nut chewers (p = 0.02), and had higher plasma B12 (p = 0.06), lower eGFR (p = 0.04), and higher systolic (p = 0.06) and diastolic blood pressure (p = 0.02) than participants having a less oxidized plasma E_hGSH (≤ −99.14). Of these factors, only age and plasma B12 were significantly independent predictors of continuous plasma E_hGSH, together explaining 5.5% of the variation in E_hGSH.

Table 1.

General characteristics of the study population.

	Total sample (N = 374; mean±SD or %)	Plasma EhGSH ≤ −99.14 (N = 187)		Plasma EhGSH > −99.14 (N = 187)		pa
		Mean±SD or %	Range	Mean±SD or %	Range
Age	43±8	42±8	30, 60	44±9	31, 63	0.05
Male (%)	48.4	44.4		52.4		0.12
BMI (kg/m2)	20.4±3.5	20.2±3.3	14.2, 31.9	20.7±3.7	13.8, 35.3	0.21
Education (years)	3.4±3.6	3.2±3.3	0, 16	3.6±3.9	0, 14	0.79
Smokers (ever) (%)	36.1	35.3		36.9		0.75
Betel nut use (ever) (%)	42.2	36.4		48.1		0.02
Own TV (%)	58.3	55.1		61.5		0.21
Water arsenic (μg/L)	138±124	146±132	0.4, 700	130±115	0.4, 457	0.32
Urinary As (μg/L)	205±236	201±219	3.7, 1340	210±253	2.3, 1800	0.79
Urinary As (SG adjusted)	232±205	250±233	13, 1088	215±171	13, 742	0.45
Blood arsenic (μg/L)	13±10	14±10	1.6, 57	13±9.1	1.2, 44	0.56
Plasma folateb (nmol/L)	13±7.2	13±7.9	2.4, 61	13±6.5	3.4, 48	0.42
Plasma B12c (pmol/L)	203±113	190±89	47, 490	216±132	44, 1183	0.06
Plasma tHcysd (μmol/L)	11±13	12±16	3, 170	11±10	3, 120	0.39
Plasma cystatin C (ng/ml)	906±258	871±224	454, 1800	941±284	447, 2000	0.04
eGFR (ml/min/1.73 m2)	94±25	97±23	35, 144	91±26	32, 142	0.04
Proteinuriae (%)	1.6	2.2		1.1		0.43
Systolic blood pressuree (mm Hg)	108±14	107±11	80, 140	110±15	83, 165	0.06
Diastolic blood pressuree (mm Hg)	72±9	71±8	48, 98	73±9	50, 100	0.02
Plasma C-reactive protein (mg/L)	2.3±4.4	2.6±5.4	0.036, 46	2.0±3.3	0.017, 32	0.50
Plasma α-1 acid glycoprotein (μg/ml)	558±192	541±180	233, 1570	575±203	205, 1340	0.09
Plasma GSH (μmol/L)	2.6±0.7	3.1±0.6	2.0, 5.5	2.1±0.5	1.0, 3.4	<0.0001
Plasma GSSG (μmol/L)	2.1±0.6	2.0±0.6	0.8, 4.7	2.2±0.6	1.0, 4.0	0.001
Plasma EhGSH (mV)	−98.7±7.3	−104.2±3.9	−116.1, −99.2	−93.1±5.1	−99.1, −73.7	<0.0001

^ap for difference between E_hGSH strata from Wilcoxon rank-sum test for continuous variables and χ² test for categorical variables.

^bN=373.

^cN=367.

^dTotal homocysteine.

^eN=368.

We calculated Spearman correlations between plasma CRP, plasma AGP, eGFR, plasma GSH, GSSG, and E_hGSH, adjusting for age, sex, and BMI. CRP and AGP were positively correlated with each other (r = 0.42, p < 0.0001), and AGP was negatively correlated with eGFR (r = −0.16, p = 0.002). CRP was negatively associated with eGFR as well; however, this correlation was not significant (r = −0.07, p = 0.21). Plasma GSH, GSSG, and E_hGSH were not correlated with CRP, AGP, or eGFR (|r| ≤ 0.05, p ≥ 0.37). Additionally, plasma GSH, GSSG, and E_hGSH were not significantly correlated with any of the As exposure measures (|r| ≤ 0.08, p ≥ 0.13), adjusting for sex, BMI, smoking, and betel nut use.

Water As, uAs-SG, and bAs were all positive predictors of plasma CRP, adjusting for age, sex, and BMI (Table 2). A 10% increase in wAs, uAs-SG, or bAs was associated with a 0.74% (p = 0.01), 0.90% (p = 0.16), and 1.39% (p = 0.07) increase in CRP, respectively. The As exposure measures were not significant predictors of plasma AGP. We observed that uAs-SG and bAs were associated with decreased eGFR, adjusting for age, sex, BMI, and smoking, although these associations did not reach statistical significance at p < 0.05; wAs was not associated with eGFR. A 10% increase in uAs-SG or bAs was associated with an average reduction in eGFR of 0.16 ml/min/1.73 m² (p = 0.12) and 0.21 ml/min/1.73 m² (p = 0.08), respectively. Adjustment for CRP attenuated the effect of uAs-SG on eGFR by 4.0% and the effect of bAs on eGFR by 4.2%. Adjustment for AGP did not attenuate the effect of uAs-SG or bAs on eGFR.

Table 2.

Covariate adjusted effect-size estimates for associations between measures of As exposure and CRP, AGP, and eGFR, in the total sample and stratified by plasma E_hGSH.

Predictor	CRPa			AGPa			eGFRb		ΔR2 (%)
	Percentage change (95% CI)	p	ΔR2 (%)c	Percentage change (95% CI)	p	ΔR2 (%)	Mean change (95% CI)	p
Total sample (N = 374)
Water As	0.74 (0.16, 1.32)	0.01	1.44	−0.01 (−0.16, 0.14)	0.86	0.01	−0.03 (−0.12, 0.07)	0.56	0.06
Urinary As-SGd	0.90 (−0.37, 2.19)	0.16	0.45	−0.04 (−0.37, 0.29)	0.82	0.01	−0.16 (−0.37, 0.04)	0.12	0.42
Blood As	1.39 (−0.10, 2.90)	0.07	0.77	−0.12 (−0.50, 0.27)	0.55	0.09	−0.21 (−0.45, 0.03)	0.08	0.54
EhGSH ≤ −99.14 mV (N = 187)
Water As	0.17 (−0.71, 1.04)	0.71	0.06	−0.06 (−0.27, 0.14)	0.55	0.18	−0.05 (−0.19, 0.09)	0.47	0.22
Urinary As-SG	−0.52 (−2.33, 1.31)	0.57	0.14	−0.26 (−0.69, 0.17)	0.24	0.69	−0.10 (−0.38, 0.19)	0.50	0.19
Blood As	−0.25 (−2.42, 1.96)	0.82	0.02	−0.35 (−0.87, 0.16)	0.18	0.91	−0.16 (−0.50, 0.18)	0.35	0.36
EhGSH > −99.14 mV (N = 187)
Water As	1.13 (0.37, 1.90)	0.004	3.93	0.03 (−0.20, 0.25)	0.82	0.03	−0.02 (−0.15, 0.11)	0.80	0.02
Urinary As-SG	2.31 (0.55, 4.10)	0.01	3.10	0.19 (−0.32, 0.71)	0.46	0.28	−0.26 (−0.56, 0.04)	0.09	0.92
Blood As	2.76 (0.75, 4.80)	0.007	3.38	0.09 (−0.49, 0.68)	0.75	0.05	−0.31 (−0.65, 0.03)	0.07	1.01

^aModels with the outcomes CRP and AGP are adjusted for log(age), sex, and log(BMI); predictor parameters are expressed as the expected % change in the geometric mean of CRP or AGP for a 10% increase in exposure.

^bModels with the outcome eGFR are adjusted for log(age), sex, log(BMI), and ever smoking; predictor parameters are expressed as the expected mean change in eGFR (ml/min/1.73 m²) for a 10% increase in exposure.

^cThe change in R² after adding As exposure variable to model.

^dUrinary As adjusted for specific gravity.

To examine modification of the effect of As exposure on CRP by plasma E_hGSH, we repeated the regression analyses within strata of plasma E_hGSH above or below the median (E_hGSH = −99.14), representing participants having relatively more or less oxidized plasma redox status, respectively (Table 2). The effect of As exposure on CRP was only significant in the higher E_hGSH strata, in which a 10% increase in wAs, uAs-SG, or bAs was associated with a 1.13% (p = 0.004), 2.31% (p = 0.01), and 2.76% (p = 0.007) increase in CRP, respectively. The regression coefficient estimates for the effect of As exposure on CRP differed between the E_hGSH strata by the Wald test, particularly for uAs-SG (p = 0.03) and bAs (p = 0.05) and less so for wAs (p = 0.10). In a sensitivity analysis, we examined the effect of As exposure on CRP in participants above and below E_hGSH cut-points ranging from the 20th to the 80th percentile of plasma E_hGSH. We found that As exposure was associated with increased CRP in groups of participants having relatively more oxidized plasma redox status (E_hGSH > cut-point), with regression coefficients and 95% CIs above zero for a wide range of E_hGSH cut-points (Fig. 1). In contrast, As exposure was not associated with CRP in groups of participants having relatively less oxidized plasma redox status (E_hGSH < cut-point); the regression coefficients had estimates around zero and the 95% CIs included zero for all E_hGSH cut-points examined. Additionally, we found that there was a range of E_hGSH cut-points with small p values from the Wald test, indicating significant differences in the effect of As exposure on CRP between groups of participants having relatively more and less oxidized plasma redox status defined by these cut-points (Fig. 2). The range of cut-points resulting in significant group differences varied by the measure of As exposure, with uAs-SG having more cut-points with small p values.

Fig. 1.

Plots of the estimated regression coefficients (B) and 95% CIs for the effect of As exposure (top, water As; middle, urinary As-SG; bottom, blood As) on log(CRP) in more and less oxidized groups, defined by different plasma E_hGSH cut-points. All models were adjusted for log(age), log(BMI), and sex.

Fig. 2.

Plots of p values from Wald tests for the difference in the effect of log(water As) (top), log(urinary As-SG) (middle), or log(blood As) (bottom) on log(CRP) between more and less oxidized groups, defined by different plasma E_hGSH cut-points. The effect of As exposure was the regression coefficient of the As variables adjusted for log (age), log(BMI), and sex.

We also found that plasma GSH concentrations modified the effect of As exposure on CRP; As exposure was associated with increased CRP (95% CIs above zero) only in participants with relatively lower plasma GSH, and there were significant differences in the effect of As exposure on CRP between plasma GSH strata by the Wald test for a wide range of plasma GSH cut-points (data not shown). Plasma GSSG concentration did not significantly modify the effect of As exposure on CRP, although significant positive associations between As exposure and CRP were observed only in participants with relatively higher plasma GSSG (data not shown).

We did not find any effect of As exposure on AGP within E_hGSH strata, defined by the median E_hGSH (Table 2) or by any other E_hGSH cut-point, nor did we find any effect of As exposure on AGP in strata defined by plasma GSH or GSSG concentrations (data not shown). For the outcome of eGFR, the negative effects of uAs-SG and bAs (but not wAs) tended to be larger in magnitude in participants with a more oxidized plasma E_hGSH (> median), compared to participants with a less oxidized E_hGSH (< median), with marginally significant regression coefficients for uAs-SG and bAs in the high E_hGSH strata (Table 2). However, the regression coefficients for the effect of As exposure on eGFR did not differ significantly between E_hGSH strata by the Wald test. Consistent results were seen in the E_hGSH cut-point sensitivity analysis for eGFR (data not shown). Likewise, we did not observe significant modification of the effect of As exposure on eGFR when stratifying by plasma GSH or GSSG concentrations (data not shown).

3. Discussion

In this study of Bangladeshi adults chronically exposed to a wide range of As concentrations through drinking water, As exposure was associated with increased plasma CRP, particularly in participants with lower plasma GSH or a relatively more oxidized plasma E_hGSH, but had no effect on plasma AGP. Our finding that As exposure was associated with increased CRP is in agreement with another study in Bangladesh [12] and also supports the recent evidence that As exposure is associated with risk for CVD [37], as CRP is a strong predictor of CVD. The finding that As exposure is associated with increased CRP in a relatively healthy population has many future health implications, as CRP is probably not only a biomarker of inflammation, but an active agent in disease pathogenesis, as has been observed for atherosclerosis [38] and insulin-related diseases [39].

CRP and AGP are both acute-phase proteins, and it is not clear why we did not observe an effect of As exposure on AGP. A potential explanation lies in the different transcription factors required for CRP and AGP activation during the acute-phase response. NF-κB family proteins, STAT3, and C/EBPβ bind to the CRP promoter to activate CRP transcription in response to IL-6 and IL-1β [40,41], whereas AGP expression in response to IL-6 involves binding of C/EBPβ and the glucocorticoid receptor to the AGP promoter [42]. Because arsenite can bind to and inhibit the glucocorticoid receptor [43], this could potentially be a reason we observe no net effect of As exposure on plasma AGP. Alternatively, the explanation may relate to differences in magnitude and duration of CRP and AGP expression during the acute-phase response. Plasma CRP increases 1000- to 10,000-fold quickly and relatively briefly after inflammatory insult, whereas AGP increases 2- to 5-fold and remains elevated longer than CRP [44]. It is possible that in a situation of chronic exposure to As, CRP is more apt to reflect the effect of variation in As exposure than AGP. Also, AGP has a much smaller range than CRP, which may hinder detection of an association.

Arsenic has been suggested to cause oxidative stress through several different pathways, including (a) increasing ROS within tissues via various proposed mechanisms [14], (b) binding to thiol groups and inhibiting regulatory proteins [45], or (c) binding directly to GSH. In pathway (a), ROS-scavenging antioxidants (e.g., vitamin C, vitamin E) would protect against the effect of As; in pathways (b) and (c), thiol-containing antioxidants (e.g., GSH, NAC, α-lipoic acid) would be necessary to increase thiol status and protect against the effect of As. Arsenic exposure has not been directly associated with oxidative stress in the FOX study [27–29], and we confirm here that As exposure is not associated with plasma GSH biomarkers. This observation is consistent with a recent study in mini pigs showing that plasma GSH biomarkers were unresponsive to oxidative stress induced by lipopolysaccharide exposure [46]. However, plasma GSH biomarkers may still reflect inherent interindividual variability in the pro-oxidant to antioxidant balance of tissues [30]. This interindividual variability may result from inheritance [47], age [48], diet [49], and other factors unrelated to As exposure. We hypothesized that participants having a more oxidized plasma E_hGSH would be at greater risk for As-induced inflammation. Indeed, we observed here that the As-induced increase in plasma CRP was apparent only in participants having relatively lower plasma GSH or a relatively more oxidized E_hGSH. The observations that plasma GSH and E_hGSH modify the effect of As exposure on CRP may suggest that higher GSH and/or greater antioxidant capacity are important in protecting against As-induced inflammation.

To our knowledge, the specific mechanism for induction of CRP expression by As has been investigated in only one study by Druwe et al. [8]. In that study, low-level arsenite exposure to HepG2 cells increased CRP expression and secretion, whereas ROS remained unchanged [8], suggesting that As can induce CRP independent of ROS. As-induced activation of NF-κB, a transcription factor involved in CRP expression [41], was prevented by pretreatment with NAC in cultured aortic endothelial cells, suggesting that oxidation of thiols was responsible for the activation of NF-κB [19]. Other transcription factors involved in CRP expression, including C/EBPβ and STAT3 [40], may also be activated by As [50,51]. Because the redox environment can modulate activation of NF-κB [52], STAT3 [53,54], and possibly C/EBPβ [55], it is possible that in a low GSH/more oxidized redox environment, As-induced redox changes mediate the effect of As on CRP, in accordance with our results. Additional research is necessary to determine the exact mechanism by which As may induce CRP expression and whether the effect of As on CRP can be inhibited by ROS-scavenging and/or thiol-containing antioxidants.

Alternatively, our findings may relate to the involvement of GSH in biliary excretion of As. Arsenite and monomethylarsonous acid form conjugates with GSH (As^III(GS)₃ and MMA^III(GS)₂), which are excreted in the bile of rats via the multidrug-resistant protein efflux transporters [56]. Additionally, arsenite and selenite enhance each other’s elimination by forming a complex with GSH ([(GS)₂AsSe]⁻) which is excreted in bile [56]. Whereas current understanding of the involvement of GSH in As excretion in humans is limited, it is possible that participants with lower GSH may have impaired biliary As elimination, resulting in greater susceptibility to As-induced inflammation. However, the primary route of As excretion is through urine, so the magnitude of such an effect is unknown.

In the present study, As exposure, as measured by As in urine and blood, was associated with reduced eGFR, although these associations did not reach statistical significance at p < 0.05. The finding that uAs, but not wAs, was negatively associated with eGFR is consistent with our previous study [4] and with other studies that have observed negative associations between uAs and eGFR [57,58]. The lack of a significant effect of As exposure on eGFR in the current study may be related to the study design, as individuals with known CKD, CVD, or diabetes were excluded from participation; these exclusion criteria would reduce the variation in eGFR in the study population. Although adjustment for CRP resulted in a very small attenuation (<5%) of the negative effect of uAs-SG and bAs on eGFR, we cannot rule out the possibility that As-induced inflammation may mediate the effects of As on renal function, because (a) the effect of As exposure on eGFR in this study was very small and (b) plasma CRP is a systemic marker of inflammation largely produced in the liver and, as such, may not specifically reflect inflammation within other tissues. Although we tested here whether inflammation may be on the causal pathway from As exposure to renal dysfunction, it is important to note that inflammation may be positively associated with cystatin C, our biomarker of GFR, independent of actual GFR [59,60]. As such, we cannot distinguish in this study whether inflammation and cystatin C-based eGFR are related through an inflammation-induced decrease in GFR or through a mechanism independent of actual GFR.

Although the negative effect of As exposure on eGFR was larger in magnitude in participants with a more oxidized plasma E_hGSH, the effect of As exposure on eGFR did not significantly differ between low and high plasma E_hGSH strata. Antioxidants are protective against As-induced renal injury and dysfunction in several animal studies [23–26]. In a human study, low levels of the antioxidant lycopene combined with high uAs were associated with increased odds of CKD, above the odds associated with either factor alone [58]. The mechanism for As-induced renal dysfunction may involve oxidative stress, however, we may not have had sufficient power to detect this in the current study. It is also possible that other biomarkers of renal injury, such as markers of proximal tubule damage, may be more appropriate for assessing As-induced nephrotoxicity.

4. Conclusions

In summary, we observed that As exposure was associated with increased CRP. Upon stratifying our analyses by plasma E_hGSH or GSH, we observed that the As-induced increase in CRP was apparent only in participants with a relatively more oxidized E_hGSH or a lower GSH concentration, suggesting that increased GSH or a higher antioxidant to pro-oxidant balance may protect against As-induced inflammation. Our conclusions are limited by the cross-sectional nature of this study, as we cannot rule out reverse causality or account for potential biases due to unmeasured confounding. Additionally, the E_hGSH strata differed on variables such as age and plasma B12 and probably differ on other unmeasured variables. Therefore, our E_hGSH strata may represent more than antioxidant capacity alone (although the majority of variation in E_hGSH in this study was not explained by measured variables). Randomized controlled trials in As-exposed humans are necessary to confirm the potential benefit of antioxidants against As-induced inflammation. Because inflammation is involved in the pathogenesis of many diseases, antioxidant treatment may represent a viable option for preventing As-induced diseases.

Highlights.

• Arsenic exposure may cause oxidative stress, inflammation, and renal dysfunction.

• We analyzed plasma GSH, GSSG, CRP, AGP, and cystatin C in 374 As-exposed adults.

• Arsenic exposure was associated with significantly increased CRP.

• Plasma GSH and E_hGSH significantly modified the effect of arsenic on CRP.

• Higher GSH/lower E_hGSH may protect against As-induced inflammation.

Abbreviations

AGP

α-1 acid glycoprotein

bAs

blood arsenic

BMI

body mass index

CKD

chronic kidney disease

CRP

C-reactive protein

CVD

cardiovascular disease

eGFR

estimated glomerular filtration rate

E_hGSH

reduction potential of the plasma GSSG/2GSH redox pair

FOX

Folate and Oxidative Stress study

GFR

glomerular filtration rate

GSH

glutathione

GSSG

glutathione disulfide

NF-κB

nuclear factor κB

ROS

reactive oxygen species

uAs

urinary arsenic

uAs-SG

urinary arsenic adjusted for specific gravity

wAs

water arsenic