Environmental Cadmium Exposure Induces an Increase in Systolic Blood Pressure by Its Effect on GFR

: Chronic exposure to the nephrotoxic metal pollutant, cadmium (Cd), has been associated with hypertension, but the mechanism by which it raises blood pressure is not understood. We hypothesize that exposure to Cd reduces the glomerular filtration rate (GFR), which in turn causes a rise in blood pressure. Data were collected from 447 Thai subjects with a mean age of 51.1 years, of which 48.8% had hypertension, 15.4% had diabetes, and 6.9% had an estimated GFR (eGFR) below 60 mL/min/1.73 m 2 (low eGFR). More than half (58.8%) and 23.9% had moderate and severe tubular proteinuria, respectively. The mean blood and urinary Cd concentrations were 2.75 and 4.23 µ g/L, respectively. Doubling of body burden of Cd increased the prevalence odds ratios (POR) for low eGFR and severe tubular proteinuria 41% and 48%, respectively. The POR for hypertension rose twofold in those with blood Cd levels of 0.61–1.69 µ g/L or urinary Cd excretion levels ≥ 0.98 µ g/g creatinine. In the hypertensive group, the eGFR was inversely associated with age ( β = − 0.517), the Cd excretion rate ( β = − 0.177), and diabetes ( β = − 0.175). By mediation analysis, an increase in SBP was attributable totally to the effect of Cd on GFR. Thus, blood pressure appeared to rise as GFR fell. This finding is consistent with the well-known role of the kidney in long-term blood pressure regulation, and explains a universally high prevalence of hypertension among patients with low eGFR.


Introduction
Hypertension, indicated by systolic blood pressure (SBP) ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg, is a common risk factor for cardiovascular disease (CVD) and can be both a cause and a consequence of chronic kidney disease (CKD) [1][2][3].Concerningly, mortality from CKD rose from the 13th leading cause of death in 2000 to the 7th in 2021, and it is projected to be the 5th leading cause of years of life lost by 2040 [4,5].
Because of its ubiquitous environmental presence, exposure to the nephrotoxic metal pollutant cadmium (Cd) is also a significant risk factor for CVD [6] and CKD [7].A 75% increase in deaths from any cause among those with CKD was associated with Cd exposure even at low levels, reflected by urinary Cd levels ≥ 0.60 µg/g creatinine [8].The risk of having CVD rose 2.58-fold and 2.79-fold at blood Cd level of 1 µg/L and urinary Cd Stresses 2024, 4 excretion rate of 0.5 µg/g creatinine, respectively [6].Notably, environmental Cd exposure appeared to adversely affect females and males differently [9][10][11][12].
In the U.S. general population, an overall mean urinary Cd excretion was 0.5 µg/g creatinine, and 2.5%, 7.1%, and 16% of non-smoking women (aged ≥ 20 years) had urinary Cd levels > 1, >0.7, and >0.5 µg/g creatinine, respectively [13].In comparison, a study from Thailand reported a 22.5% prevalence of Cd excretion ≥ 1 µg/g creatinine in non-smoking women who had low body iron stores [14].Thus, the proportions of at-risk subpopulations are concerning.
Effects of environmental exposure to Cd on blood pressure has been well documented in studies of the general populations of the U.S. [15][16][17], Canada [18], China [19][20][21], Korea [22,23], and Japan [24].For non-occupationally exposed populations, diet is the main source of Cd exposure, other than through passive and active smoking [25][26][27][28].The intestinal absorption rate of Cd can be as high as 45% [29,30] as Cd can be absorbed through several metal transporters for metal nutrients, iron, zinc, copper, and calcium (Fe, Zn, Cu and Ca) [31,32].Furthermore, Cd complexed with metallothionein (MT) and phytochelatin (PC), as CdMT and CdPC, can be absorbed through transcytosis as well as receptor-mediated endocytosis [33][34][35].Notably, however, there are no physiologic mechanisms for Cd elimination.Continued exposure will lead to its accumulation in tissues and organs throughout the body, notably in the kidneys, where it promoted the progression of kidney disease to kidney failure, especially in diabetics [36].
The kidneys play an indispensable role in long-term blood pressure regulation [2], and the principal site of Cd toxicity due to a preferential accumulation of Cd in the proximal tubular epithelium [37,38].As these cells die, CdMT is released into tubular lumen and is excreted in urine [39].Most or all excreted Cd originates from injured or dying kidney tubular epithelial cells, thus the excretion of Cd itself indicates the severity of the kidney injury at the present time [39].A correlation between kidney accumulation and urinary Cd concentration also forms the basis for urinary Cd as an indicator of long-term exposure to Cd or its body burden [40].
The deleterious effects of Cd exposure on the kidneys have been investigated extensively in workplace and non-workplace exposure settings [41].However, a few studies have explored the mechanism by which Cd raises blood pressure.The present study aimed to test the hypothesis that an increase in blood pressure is the result of kidney tubular cell damage by Cd.Because a reduction in the glomerular filtration rate (GFR) is a common sequela of ischemic acute tubular necrosis, and acute and chronic tubulointerstitial fibrosis, all of which create impediments to filtration [42,43], we quantified tubular cell damage, tubular proteinuria, systolic, and diastolic blood pressures (SBP and DBP) according to the estimated GFR (eGFR) and Cd exposure levels in those diagnosed with and without hypertension.Urinary excretion of Cd (E Cd ) and blood Cd concentrations ([Cd] b ) were used as measures of exposure levels.Urinary excretion of β 2 -microglobulin (β 2 M) and N-acetyl-β-D-glucosaminidase (NAG) were used to assess tubular proteinuria and damage to the kidney tubular cells, respectively [41].

Demographic and Biologic Characteristics of Participants
A total of 447 persons (333 women and 114 men) with a mean age of 51.1 years, were recruited to this study (Table 1 [45].SBP and DBP data were from 445 subjects.Data on E NAG /E cr and (E NAG /C cr ) × 100 were from 308 subjects.All other data were from 447 subjects.Arithmetic mean ± standard deviation (SD) values are provided for all continuous variables.p ≤ 0.05 identifies statistical significance.
[Cd] u and [Cd] b correlated strongly with each other in both women and men with respective Pearson's correlation coefficients of 0.688 and 0.615.After controlling for age, [Cd] u and [Cd] b correlations remained with partial r values of 0.584 and 0.658 in women and men, respectively.
The percentages (%) of smoking in the top (42.7%) and middle (34.9%)E Cd /C cr tertile groups were higher than the low (16.2%)tertile group.The % hypertension were similar across E Cd /C cr tertiles, but diabetes was more prevalent in the low tertile (39.2%) compared with the middle (3.4%) and high (4.0%)tertile groups.In parallel, low eGFR was more prevalent in the low tertile (10.3%) than in the middle (1.3%) and top (8.7%) tertiles.
More than half (58.8%) of participants had moderate proteinuria, and 23.9% had severe proteinuria.The % severe tubular proteinuria rose to 31.3% in the top E Cd /C cr tertile, compared with 14.8% in the middle tertile group.
Mean age (56.6 years), mean BMI (25.5 kg/m 2 ), mean SBP (134 mmHg), and mean DBP (83 mmHg) in the low tertile group were all statistically higher than the middle and top E Cd /C cr tertile groups.The mean eGFR of 84 mL/min/1.73m 2 in the low tertile was 7-12 mL/min/1.73m 2 below the mean eGFR values in the middle and top tertiles.

Cadmium, Hypertension, Low eGFR and Tubular Proteinuria
We used logistic regression to evaluate associations of age, BMI, gender, smoking, diabetes, and Cd burden with prevalence odds ratios (POR) for hypertension (Table 2).The prevalence odds ratio (POR) for hypertension rose with BMI (POR 1.082, 95% CI: 1.027-1.140)and Cd burden at a medium level (POR 2.114, 95% CI: 1.049-4.260).An increase in POR for hypertension was statistically insignificantly in the heavy Cd burden group (p = 0.092).
To evaluate the effects of Cd on the prevalence of low eGFR and tubular proteinuria, three more logistic regression models were conducted (Table 4).
For every one-year older, the POR for low eGFR, moderate, and severe tubular proteinuria rose 14.6%, 3.7%, and 6.4%, while doubling of body burden of Cd was associated with 41%, 23%, and 48% increases in the POR for low eGFR, moderate, and severe tubular proteinuria, respectively.Among diabetics, there were 4.3-fold, 5.5-fold, and 13.4-fold increases in the POR for low eGFR, moderate, and severe tubular proteinuria, respectively.

Comparing Effects of Cadmium on eGFR in Women and Men
Scatterplots that related eGFR to the Cd excretion rates Cd in women and men are presented in Figure 1.For every one-year older, the POR for low eGFR, moderate, and severe tubular proteinuria rose 14.6%, 3.7%, and 6.4%, while doubling of body burden of Cd was associated with 41%, 23%, and 48% increases in the POR for low eGFR, moderate, and severe tubular proteinuria, respectively.Among diabetics, there were 4.3-fold, 5.5-fold, and 13.4-fold increases in the POR for low eGFR, moderate, and severe tubular proteinuria, respectively.

Comparing Effects of Cadmium on eGFR in Women and Men
Scatterplots that related eGFR to the Cd excretion rates Cd in women and men are presented in Figure 1.A significant inverse dose-response relationship was observed between eGFR and E Cd /C cr in women (β = −0.187)and men (β = −0.322)who had medium plus heavy Cd burdens (Figure 1b), but not in those with a mild Cd burden (Figure 1a).

Comparing Effects of Cadmium on eGFR in the Normotensive and Hypertensive Groups
Scatterplots that related eGFR to the excretion rate of Cd in the normotensive and hypertensive groups are presented in Figure 2.
Lower eGFR values were associated with higher E Cd /C cr values in both normotensive (β = −0.034)and hypertensive groups (β = −0.070)who had medium plus heavy Cd burdens (Figure 2b).In comparison, eGFR and E Cd /C cr were not corelated with each other in the groups with a mild Cd burden, regardless of blood pressure status (Figure 2a).

Comparing Effects of Cadmium on eGFR in the Normotensive and Hypertensive Groups
Scatterplots that related eGFR to the excretion rate of Cd in the normotensive and hypertensive groups are presented in Figure 2. Lower eGFR values were associated with higher ECd/Ccr values in both normotensive (β = −0.034)and hypertensive groups (β = −0.070)who had medium plus heavy Cd burdens (Figure 2b).In comparison, eGFR and ECd/Ccr were not corelated with each other in the groups with a mild Cd burden, regardless of blood pressure status (Figure 2a).

Inverse Relationships between Blood Pressure and eGFR
Scatterplots relating blood pressure levels to eGFR can be found in Figure 3.

Inverse Relationships between Blood Pressure and eGFR
Scatterplots relating blood pressure levels to eGFR can be found in Figure 3.A significant inverse dose-response relationship was evident between SBP and eGFR in women (β = −0.227)and men (β = −0.320) of the medium plus heavy Cd burden group (Figure 3b), but not a mild Cd burden group (Figure 3a).In comparison, DBP did not significantly correlate with eGFR in women or men at any Cd burden (Figure 3c,d).

Regression Model Analysis of SBP and DBP
Results of an evaluation of the independent effect of GFR and diabetes on SBP and DBP are presented in Table 6.
Age, BMI, log 2 [(E Cd /C cr ) × 10 5 ], eGFR, gender, hypertension, smoking, and diabetes together contributed, respectively, to 19.9%, 15.7%, and 15.0% of the variation in SBP in all subjects, the mild Cd burden, and medium plus heavy Cd burden groups.The fractional DBP variation explained by these independent variables in the mild Cd burden, and medium plus heavy Cd burden groups, were 4.6%, 0%, and 5.8%, respectively.Mild, medium plus heavy Cd burdens were defined as (E Cd /C cr ) × 100 < 1 and ≥ 1 µg/L, respectively.eGFR, estimated glomerular filtration rate; β, standardized regression coefficient; adjusted R 2 , coefficient of determination.β indicates strength of association of eGFR with seven independent variables (first column).Adjusted R 2 indicates the fraction of SBP and DBP variability that all independent variables explained.p-values ≤ 0.05 indicate statistically significant associations of independent variables and blood pressure.

Mediation Analysis
We employed a simple mediation model analysis to explore whether Cd increased blood pressure through its effect on GFR.In this analysis, eGFR was a single mediator, while E Cd /C cr was an independent variable, and SBP or DBP was a dependent variable).Results for those with (E Cd /C cr ) ×100 ≥ 1 µg/L filtrate (the high Cd-group, n = 322) can be found in Figure 4.
In the high-Cd group, Cd did not affect SBP or DBP directly, but through its effect on GFR.GFR was a full mediator of the increment of both SBP and DBP in Cd exposed persons with (E Cd /C cr ) ×100 ≥ 1 µg/L filtrate.
An equivalent mediation analysis was undertaken for those with (E Cd /C cr ) ×100 < 1 µg/L filtrate (the low-Cd burden group, n = 123), and results are provided in Figure 5.
In the low-Cd group (Figure 5), an effect of eGFR on SBP was statistically significant (standardized β = −0.186,p = 0.041).However, the Sobel test result (A*B) indicated that eGFR was not a mediator of the effect of Cd on SBP (p = 0.309) or DBP (p = 0.595).

Mediation Analysis
We employed a simple mediation model analysis to explore whether Cd increased blood pressure through its effect on GFR.In this analysis, eGFR was a single mediator, while ECd/Ccr was an independent variable, and SBP or DBP was a dependent variable).Results for those with (ECd/Ccr) ×100 ≥ 1 µg/L filtrate (the high Cd-group, n = 322) can be found in Figure 4.In the high-Cd group, Cd did not affect SBP or DBP directly, but through its effect on GFR.GFR was a full mediator of the increment of both SBP and DBP in Cd exposed persons with (ECd/Ccr) ×100 ≥ 1 µg/L filtrate.
An equivalent mediation analysis was undertaken for those with (ECd/Ccr) ×100 < 1 µg/L filtrate (the low-Cd burden group, n = 123), and results are provided in Figure 5.  found in Figure 4.In the high-Cd group, Cd did not affect SBP or DBP directly, but through its effect on GFR.GFR was a full mediator of the increment of both SBP and DBP in Cd exposed persons with (ECd/Ccr) ×100 ≥ 1 µg/L filtrate.
An equivalent mediation analysis was undertaken for those with (ECd/Ccr) ×100 < 1 µg/L filtrate (the low-Cd burden group, n = 123), and results are provided in Figure 5.

Discussion
In this Thai cohort of 447 adults, the overall mean Cd excretion rate was 0.003 µg/L filtrate, corresponding to 4.03 µg/g creatinine.The mean blood and urinary Cd concentrations were 2.75 and 4.23 µg/L, respectively.Nearly half (48.8%) of the cohort's participants had hypertension, while 15.4% and 6.9% had diabetes and low eGFR, respectively.Similarly, a study from Bangladesh reported that half of those aged ≥ 60 years had hypertension [46].The prevalence of low eGFR in our study was in line with the figure of 6.3% found in the Taiwanese general population [47].However, the % of hypertension and diabetes in the present study were higher than those recorded in studies from the U.S., where hypertension and diabetes were 39% and 10.3-13%, respectively [48,49].
The high prevalence of hypertension in this cohort was its strength as it meant that even a modest sample size (n < 1000) could offer a sufficient number of cases from which a reliable conclusion could be drawn.Previously, an effect of smoking on the risk of CVD had been found to be partially mediated by Cd [50,51].An inclusion of smokers and diabetics was an additional strength as they enabled an adjustment for their effects in realistic population situations.Another strength was that Cd exposure was assessed with blood and urinary Cd levels.
The limitations of the study are acknowledged.They include a one-time-only assessment of Cd exposure and its effects, and the heterogeneity in the hormonal status of women; both menopausal and post-menopausal women were included [52][53][54][55][56], and there was a small number of men (n = 114) in the study cohort; this meant that definitive conclusions about gender disparity in the prevalence, severity, and adverse outcomes of Cd-induced hypertension could not be made [53][54][55][56].

A Rise in Blood Pressure at Low Levels of Cadmium Exposure
An increased risk of hypertension was associated with both urinary and blood Cd levels (Tables 2 and 3).POR for hypertension rose two-fold in the medium burden group and in those with the blood Cd quartiles 2 and 3.However, the increases in POR for hypertension in those with a heavy burden of Cd (POR 1.66) and those with the top blood Cd quartile (POR 1.80) were not statistically significant.Thus, Cd effects on blood pressure appeared to be particularly strong in low-dose exposure conditions as detailed below.
In a Chinese case-control study, a 1.33-fold increase in the risk of hypertension was associated with urinary Cd levels > 1.07 µg/L [20].In the present study, a two-fold rise in POR for hypertension was observed in the medium burden group with urinary Cd levels > 0.57 µg/L (Table 2).A 2.6-fold increase in risk of hypertension was seen in white and Mexican-American women who had blood Cd levels ≥ 0.4 µg/L [15].In the present study, an increase in POR for hypertension was found in those who had blood Cd of 0.61-3.38µg/L (Table 3).
In a study from Korea, increases in prevalence of pre-hypertension and hypertension were associated with doubling blood Cd from 0.62 to 1.33 µg/L in men, and from 0.73 to 1.57 µg/L in women [22].In a study of residents in a Cd-polluted area of China, increased risk of hypertension was associated with blood Cd of 1-1.7 µg/L [19].
In a Canadian study, SBP and DBP were positively associated with blood Cd, but the risk of hypertension fell 52% in women who were current smokers and had a high blood Cd level [18].In the Canadian health measure survey, the mean values for blood and urinary excretion of Cd in current smokers were 1.64 µg/L and 0.58 µg/g creatinine, respectively [57].A similar observation was made in a U.S. population study, where associations between blood pressure measurements and blood Cd were particularly strong in non-smokers, moderate in former smokers, and weak or negligible in current smokers [15].

Different Susceptibility to Cadmium-Induced Hypertension
The urinary and blood Cd levels found to be associated with a significant increase in risk of hypertension varied among populations.This may be due to different susceptibility to hypertension or some protective factors.For instance, white and Mexican-American women were found to be more susceptible to Cd-induced blood pressure increases than black women; an increased risk of hypertension was seen in Caucasian (OR 1.54) and Mexican-American women (OR 2.38) who had blood Cd as little as 0.4 µg/L, but not in black women or white, black, or Mexican-American men [16].
Male-female differences were evident from a regression model analysis (Table 5), where an inverse association of eGFR and Cd burden was found only in women (β = −0.121,p = 0.051).In comparison, male eGFR did not show a significant association with Cd burden (β = −0.077,p = 0.463).In a Taiwanese study, an association of urinary Cd and a tubular damage marker (urinary NAG) was found in women only [58].In a Thai population study, the risk of hypertension rose 20% in Cd-exposed subjects with kidney tubular damage, assessed with urinary NAG excretion [59].
People with diabetes were more susceptible to adverse kidney effects of Cd than their non-diabetic counterparts.In a Dutch cross-sectional study, including 231 patients with type 2 diabetes, Cd exposure increased the risk of diabetic kidney disease [60].In a six-year median follow-up of these diabetic patients, a progressive reduction of eGFR was attributable to Cd exposure [36].
In a prospective cohort study, a 49% increase in all-cause mortality among the diabetics was associated with urinary Cd levels > 0.60 µg/L [61].In the present study, the risk of having low eGFR, moderate, and severe tubular proteinuria among diabetics rose 4.3-fold, 5.5-fold, and 13.4-fold, respectively (Table 4).Both SBP (β = 0.265) and DBP (β = 0.193) rose significantly among diabetics with (E Cd /C cr ) × 100 ≥ 1 µg/L filtrate (Table 6).This Cdinduced SBP and DBP increment may promote kidney disease development in Cd-exposed diabetics given that hypertension is a strong independent risk factor for the development and progression of CKD [3,62].A 1.76-fold increase in death from CVD among U.S. citizens with hypertension was associated with elevated Cd exposure, indicated by blood Cd levels ≥ 0.80 µg/L [63].At the same Cd exposure level, the risk of death from CVD rose 2.12-fold among non-smokers who had hypertension [63].

A Rise of Blood Pressure Due to Tubular Damage and GFR Loss
To the best of our knowledge, the present study has provided, for the first time, evidence linking Cd-induced eGFR reduction to a rise in blood pressure (Figure 3).SBP was inversely associated with eGFR in women (β = −0.227)and men (β = −0.320)who had medium plus heavy Cd burdens, (E Cd /C cr ) × 100 ≥ 1 µg/L filtrate.DBP showed a weak inverse association with eGFR (Figure 3a,b vs. Figure 3c,d).By multiple regression analysis (Table 6), an independent effect of eGFR reduction on a rise of SBP was found in the medium plus heavy Cd burden group (β = −0.176).Using the mediation analysis, GFR reduction was a full mediator of Cd effect on blood pressure increases in the high-exposure group (Figure 4).Our observation that blood pressure rises as GFR falls helps to explain why patients with CKD have hypertension almost universally.
The doubling of the body burden of Cd was associated, respectively, with increases of 41%, 23%, and 48% in the POR values for low eGFR, moderate, and severe tubular proteinuria, evident from increased β 2 M excretion levels (Table 4).In Japanese population studies, increased β 2 M excretion levels were associated with enhanced risks of hypertension and a large decline in eGFR (10 mL/min/1.73m 2 ) over a five-year observation period [64,65].Thus, an increased risk of hypertension among study subjects could be attributed to Cdinduced tubular damage and GFR loss.
In summary, we have shown that increases in blood pressure may be a consequence of a decrease in GFR induced by Cd.The indispensable role of the kidneys in long-term blood pressure regulation is well established [2].As their function declines (indicated by low eGFR), the kidneys eliminate less water and sodium, which may increase blood pressure.Rats with Cd-induced hypertension showed increased sodium retention and reduced sodium excretion [66][67][68].Thus, increased tubular avidity for filtered sodium appeared to be a possible mechanism by which lifelong, low-dose Cd intake caused hypertension.
CKD has now reached epidemic proportions and is predicted to become an even greater health problem in years to come as its major risk factors-obesity, diabetes, hypertension, and non-alcoholic fatty liver-continue to rise.Given the immense financial and community burden, developing strategies to reduce its progression to kidney failure is of vital importance.Cd excretion corresponding to a discernible GFR decline at E Cd /C cr of 0.01 µg/L filtrate is extremely low.

Participants
This study was conducted following the principles outlined in the Declaration of Helsinki.Participants were recruited from Nakhon Si Thammarat Province in the south and Mae Sot District in the northwest of Thailand [69,70].Previous studies suggest female preponderance effects Cd exposure [9,12].Thus, more women were recruited to maximize the likelihood of finding an effect of Cd in a modest sample size (n = 447).
All participants gave informed consent prior to participation.They had been living at their current addresses for at least 30 years.Exclusion criteria were pregnancy, breastfeeding, a history of metal work, and a hospital record or physician's diagnosis of an advanced chronic disease.All subjects were provided with details of study objectives, study procedures, benefits, and potential risks, and they all provided their written informed consent prior to participation.
The Office of the Human Research Ethics Committee of Walailak University approved the study protocol for the Nakhon Si Thammarat group (Approval number WUEC-20-132-01, 28 May 2020) [69].The study protocol for the Mae Sot group was approved by the Institutional Ethical Committees of Chiang Mai University and the Mae Sot Hospital (Approval No. 142/2544, 5 October 2001) [70].
Levels of various contaminants, including arsenic, chromium, lead, and Cd in samples of soils and food crops in Nakhon Si Thammarat were within permissible ranges [68], and no association was found between water arsenic concentration and the risk of diabetes [71,72].In comparison, the Cd concentration of the paddy soil samples from the Mae Sot district exceeded the standard of 0.15 mg/kg, and the rice samples collected from household storage contained four times the amount of the permissible Cd level of 0.1 mg/kg [73].In a health survey of residents of the Mae Sot District (n = 5273), urinary Cd excretion levels correlated with hypertension and diabetes [74].

Blood Pressure and Cadmium Exposure Ascertainment
Our study design was population-based and recruited participants from their communities.This precluded a 24 h measurement, and more visits to communities for a second measurement.The one-time measurement of SBP/DBP was the average of 3 repeated measurements.The diagnosis of hypertension relied primarily on the assessment made by the presiding physician and the recorded use of anti-hypertensive medication.Of 233 subjects included as hypertensive cases in our study, 220 (94.4%) were being treated and 13 hypertensive cases were identified during our visit.
Cd exposure was based on urinary Cd excretion (E Cd ) and blood Cd concentration ([Cd] b ).Simultaneous urine and whole blood sampling were undertaken after an overnight fast.Aliquots of blood and urine samples were stored at −80 • C for later analysis.Atomic absorption spectrophotometry was used to determine urinary and blood levels of [Cd] u and [Cd] b using multi-element standards (Merck KGaA, Darmstadt, Germany) for instrument calibration.For quality control and assurance purposes, blood and urine samples from subjects, blood control samples (ClinChek, Munich, Germany), and the reference urine metal controls (Lyphocheck, Bio-Rad, Hercules, CA, USA) were simultaneously analyzed.
The limit of detection (LOD) for Cd in blood or urine was 0.3 µg/L for [Cd] b and 0.1 µg/L for [Cd] u .The Cd concentration assigned to a sample that contained Cd below its LOD was assigned a value of the LOD divided by the square root of 2 [75].).E Cd /C cr was expressed as an amount of Cd excreted per volume of the glomerular filtrate.C cr -normalization corrects for urine dilution and number of functioning nephrons [45].

Normalization of Cadmium Excretion Rate
E Cd was normalized to E cr as [Cd] u /[cr] u , where [Cd] u = urine concentration of Cd (mass/volume) and [cr] u = urine creatinine concentration (mg/dL).E Cd /E cr was expressed in µg/g creatinine.E cr -normalization corrects for urine dilution, but it is influenced by muscle mass.The effect of Cd exposure on GFR was obscure, when E Cd was normalized to E cr [76].

Estimated Glomerular Filtration Rate (eGFR)
We used the GFR estimating equations, established by the chronic kidney disease epidemiology collaboration (CKD-EPI) to compute the estimated GFR (eGFR) [44].The CKD-EPI equations have been validated with inulin clearance [77]. Male

Statistical Analysis
Data were analyzed with IBM SPSS Statistics 21 (IBM Inc., New York, NY, USA).The Kruskal-Wallis test was used to assess differences in means across tertiles of Cd burden, and the Pearson chi-squared test was used to assess differences in percentages.Distribution of continuous variables was assessed by the one-sample Kolmogorov-Smirnov test.A logarithmic transformation was applied to variables that showed rightward skewing.A simple mediation model with a single mediator was used in the mediation analysis [78,79] Prevalence Odds Ratio (POR) for hypertension, which was defined as SBP ≥ 140 mmHg or DBP ≥ 90 mmHg [1], was determined by logistic regression.Multiple linear regression was used to identify variables affecting eGFR, SBP, and DBP.For all tests, p-values ≤ 0.05 were considered as statistically significant.

Conclusions
By mediation analysis, an increase in SBP was attributable totally to Cd-induced GFR loss.Cd appeared to influence SBP more markedly than DBP.A two-fold increased risk of hypertension was associated with urinary Cd excretion of 0.98 µg/g creatinine and a blood Cd level of 0.61 µg/L.As these are levels that are reported widely in studies of non-occupationally exposed populations from across the world, it is imperative that authorities monitor the environmental levels of Cd closely, especially in staple foods.

Figure 1 .
Figure 1.Comparing cadmium effects on GFR in women and men.Scatterplots relate eGFR to log[(E Cd /C cr ) × 10 5 ] in women and men with (E Cd /C cr ) × 100 < 1 µg/L filtrate (a) and E Cd /C cr ) × 100 ≥ 1 µg/L filtrate (b).Coefficients of determination (R 2 ) and standardized β-coefficients for all scatterplots, numbers of subjects in subgroups, and p-values are provided.

Figure 2 .
Figure 2. Comparing the effects of cadmium on GFR in the normotensive and hypertensive groups.Scatterplots relate eGFR to log[(ECd/Ccr) × 105] in normotensive and hypertensive with (E Cd /C cr ) × 100 < 1 µg/L filtrate (a) and E Cd /C cr ) × 100 ≥ 1 µg/L filtrate (b).Coefficients of determination (R2) and standardized β-coefficients for all scatterplots, numbers of subjects in subgroups, and p-values are provided.

Figure 3 .
Figure 3. Cadmium and eGFR as predictors of blood pressure increases.Scatterplots relate SBP (a,b) and DBP (c,d) to eGFR in women and men with (ECd/Ccr) × 100 of <1 and ≥ 1 µg/L filtrate.Coefficients of determination (R 2 ) and standardized β-coefficients for all scatterplots, numbers of participants in subgroups, and p-values are provided.

Figure 3 .
Figure 3. Cadmium and eGFR as predictors of blood pressure increases.Scatterplots relate SBP (a,b) and DBP (c,d) to eGFR in women and men with (E Cd /C cr ) × 100 of <1 and ≥ 1 µg/L filtrate.Coefficients of determination (R 2 ) and standardized β-coefficients for all scatterplots, numbers of participants in subgroups, and p-values are provided.

Figure 4 .
Figure 4. Mediation analysis of the effect of cadmium on blood pressure in the high-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values (b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C′).

Figure 5 .
Figure 5. Mediation analysis of the effect of cadmium on blood pressure in the low-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values.(b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C′).

Figure 4 .
Figure 4. Mediation analysis of the effect of cadmium on blood pressure in the high-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values (b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C ′ ).

Figure 4 .
Figure 4. Mediation analysis of the effect of cadmium on blood pressure in the high-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values (b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C′).

Figure 5 .
Figure 5. Mediation analysis of the effect of cadmium on blood pressure in the low-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values.(b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C′).

Figure 5 .
Figure 5. Mediation analysis of the effect of cadmium on blood pressure in the low-exposure group.(a) A model depicts eGFR as a mediator of the effect of Cd on blood pressure increases and standardized β values.(b) The Sobel test of unstandardized β coefficients describing relationships of Cd with eGFR (A), eGFR with blood pressure (B), and Cd with blood pressure (C ′ ).

E
Cd was normalized to creatinine clearance (C cr ) as E Cd /C cr = [Cd] u [cr] p /[cr] u , where [Cd] u = urine concentration of Cd (mass/volume); [cr] p = plasma creatinine concentration (mg/dL); and [cr] u = urine creatinine concentration (mg/dL ). Overall mean [Cd] u and [Cd] b values were 4.23 and 2.75 µg/L, respectively.Participants were grouped by the tertile of the Cd excretion rate [(E Cd /C cr ) × 100].Corresponding mean (E Cd /C cr ) × 100 values in the low, middle, and high tertile groups were 0.38, 2.28 and 6.89 µg/L filtrate, equivalent to E Cd /E cr of 0.48, 3.07 and 8.48 µg/g creatinine.Mean blood Cd concentration [Cd] b values in the low, middle and high tertiles of Cd burden were 0.72, 2.37, and 5.14 µg/L, respectively.

Table 1 .
Descriptive characteristics of study subjects according to cadmium burden tertiles.

Table 2 .
Effects of cadmium burden on prevalence odds ratios for hypertension.

Table 3 .
Prevalence odds ratios for hypertension in relation to blood cadmium quartiles.

Table 4 .
Prevalence odds ratios for low eGFR and tubular proteinuria in relation to cadmium body burden and other independent variables.

Table 5 .
Comparing inverse associations of eGFR with the cadmium excretion rate in subjects grouped by gender and blood pressure status.

Table 5 .
Comparing inverse associations of eGFR with the cadmium excretion rate in subjects grouped by gender and blood pressure status., number of subjects; eGFR, estimated glomerular filtration rate; β, standardized regression coefficient; BMI, body mass index; adjusted R 2 , coefficient of determination.β indicates strength of association of eGFR with seven independent variables (first column).Adjusted R 2 indicates the proportion of eGFR variation explained by all independent variables.p-values ≤ 0.05 indicate statistically significant associations of independent variables and eGFR. n

Table 6 .
Multiple linear regression analysis to evaluate association of systolic and diastolic blood pressures with cadmium, eGFR, and other variables.