Diabetic Kidney Disease Associated with Chronic Exposure to Low Doses of Environmental Cadmium

1. Introduction

Prediabetes and diabetes (DM) are respectively designated when fasting plasma glucose (FPG) concentrations ≥ 110 and ≥ 126 mg/dL [1]. Persistent albuminuria and a progressive decline in an estimated glomerular filtration rate (eGFR) are the complications of diabetes, known as diabetic kidney disease (DKD) [2,3,4]. The severity of DKD has been based on albumin excretion rate (E_alb), expressed as urinary albumin-to-creatinine ratio (ACR) where microalbuminuria and macroalbuminuria are described [3,4]. Notably, however, about 30% of DKD patients had abnormal eGFR together with normal ACR figures; <20 mg/g creatinine (cr) in men and <30 mg/g cr in women [4,5,6]. A wide spectrum of DKD has increasingly been recognized, suggesting involvement of different pathogenic factors. Dietary exposure to low doses of the metal contaminant cadmium (Cd) has emerged as one of the potential contributors to DKD onset and its progression; the mechanism, however, remains unclear.

The kidney and liver are the two organs in the body that produce and release glucose into the circulation [7,8,9,10]. In addition, the kidney is responsible for filtration, and retrieval of 160 to 180 g of glucose each day, mediated by the sodium glucose co-transporter 1 (SGLT1) and SGLT2 [10]. Increased abundance of the glucose transporters may raise the renal threshold for glucose excretion [8]. The deterioration of eGFR can be attenuated by SGLT2 inhibitors [11].

Acquired Cd accumulates mostly within the kidney tubular cells, where its levels increase through to the age of 50 years but decline thereafter due to its release into the urine as the injured tubular cells die [12]. Thus, excretion of Cd (E_Cd) reflects kidney accumulation of Cd and its nephrotoxicity [13]. A prospective cohort study causally linked a fall of eGFR at a high rate to E_Cd [14]. Furthermore, increased risks of prediabetes, diabetes and albuminuria were associated with E_Cd, while FPG correlated with E_Cd [15,16,17,18,19]. Excretion of N-acetylglucosaminidase (NAG) and kidney injury molecule-1 have also been related to E_Cd [20,21,22].

Intriguingly, the excretion rates of NAG (E_NAG) and β₂-microglobulin (E_β2M) were enhanced in DM [23,24,25,26,27], while glycemic risk indices were associated with an elevated E_NAG that preceded albuminuria [28]. This study explored how Cd exposure induces albuminuria and a falling eGFR by examining associations of E_alb, E_NAG and eGFR in Cd-exposed DM (n = 65) and Cd-exposed controls (n = 72). Also, it was to show the effects of adjusting E_alb, E_NAG, E_β2M and E_Cd to creatinine excretion (E_cr) that obscured Cd effects.

2. Results

2.1. Descriptive Data on the Controls and Diabetics

Characteristics of the controls and diabetes are enlisted in Table 1.

This cohort had 30 men and 107 women with the overall mean age (range) 59.7 (41−80) years and the overall mean BMI (range) of 25.6 (15-48) kg/m². Respective percentages (%) of smoking, reduced eGFR and hypertension were 10.4, 11.9 and 54.2. Nearly half (49.3%) had FPG ≥ 110 mg/dL, while 39.6 % had FPG ≥ 126 mg/dL. The overall % E_cr-based albuminuria was 22.4 and 26 for C_cr-based albuminuria.

Across the three study groups, % smoking and reduced eGFR were similar, while % hypertension was highest, middle, and lowest in the ≥ 10-yr DM (80), < 10-yr DM (54.1) and CTRL (44.9). C_cr-based microalbuminuria was highest, middle, and lowest in the ≥ 10-yr DM (52), <10-yr DM (33.3) and CTRL (12.9), while % macroalbuminuria were 0, 10.8 and 8. The variations in age, eGFR, DBP, E_Cd/E_cr, and E_Cd/C_cr across the three groups were insignificant. The variables showing significant variations across the three study groups were BMI, FPG, SBP, [β₂M]_s, [β₂M]_u, E_alb/E_cr , and E_alb/C_cr.

2.2. Comparing Cd Exposure and other Measured Variables in Three Study Groups

The distribution of E_Cd, eGFR, FPG ([Glc]_p) and serum β₂M are presented in Figure 1.

Neither E_Cd/C_cr (Figure 1A) nor eGFR (Figure 1B) show a significant variation across the three study groups, while FPG (Figure 1C) and serum β₂M (Figure 1D) were higher in diabetics than controls.

Figure 2 presents the distribution of data on kidney tubular cell injury (E_NAG/C_cr) and its function, assessed with E_alb/C_cr, E_β2M/C_cr and FrTD_β2M.

E_NAG/C_cr (Figure 2A), E_alb/C_cr (Figure 2B) and E_β2M/C_cr (Figure 2C) were elevated in the diabetics, compared to controls, while the FrTD_β2M was reduced in the diabetics compared to controls (Figure 2D).

2.3. Logistic Regression Modeling of Albuminuria and reduced eGFR

Logistic regression model analysis was used to define the determinants of the prevalence odds ratios (POR) for albuminuria and reduced eGFR using C_cr-based data (Table 2).

Age, BMI, E_Cd/C_cr, smoking and gender had little effects on the POR for albuminuria. However, this parameter was increased 4.3-fold, 4.1-fold, and 2.8-fold in those with reduced eGFR (p = 0.027), diabetes (p = 0.020), and hypertension (p = 0.032). Age and E_β2M/C_cr ≥ 3, µg/L filtrate were associated with 1.15-fold and 6.4-fold increases in the POR for reduced eGFR, respectively.

For E_cr-based data (Table S1), the POR for albuminuria was increased 5.7-fold, 7.1-fold, and 3.7-fold in women (p = 0.015), those with diabetes (p < 0.001) and hypertension (p = 0.027). Notably, however, an association of E_cr-based albuminuria with reduced eGFR did not reach a statistically significance level (p = 0.052). Age was the only variable associated with reduced eGFR (POR 1.052, p = 0.002) (Table S2).

Results of covariance analysis of E_alb and eGFR are provided in Table 3 and Table 4.

As shown in Table 3 and Table 4, variables altogether explained significant fractions of E_alb and eGFR variability in CTRL, DM, and women, but they explain a little E_alb variation in men was insignificant. Results of covariance analyses of E_alb and eGFR using E_cr-based data are provided in Tables S3 and S4.

In CTRL (Table 3), E_alb was influenced by FrTD_β2M (η² = 0.228), hypertension (η² = 0.210). smoking (η² = 0.194) and gender (η² = 0.115). In DM, E_alb in diabetes was influenced by gender (η² = 0.176), E_NAG (η² = 0.162), hypertension (η² = 0.146), smoking (η² = 0.107), and BMI (η² = 0.097), while E_Cd and FrTD_β2M. showed minimal effects.

In CTRL (Table 4), E_NAG explained the largest fraction eGFR variation (η² = 0.301). The eGFR variation among DM was explained by E_NAG (η² = 0.199), E_β2M (η² = 0.183) and age (η² = 0.140). In women, eGFR was explained by E_NAG (η² = 0.222), FPG (η² = 0.109), and E_β2M (η² = 0.072).

2.4. Bivariate Analysis of E_alb

Table 5 provides results of the Spearman’s rank correlation analysis.

E_alb varied inversely with FrTD_β2M (r = −0.237) and directly with FPG (r = 0.273), E_NAG (r = 0.327) and E_β2M (r = 0.265). A correlation between E_alb and E_Cd was insignificant. Apparently, the strength of E_alb/E_NAG correlation was the highest, compared to other 4 variables tested. Consequently, scatterplots and regression analysis were used to further examined E_alb and E_NAG/C_cr relationship in subgroups (Figure 3).

E_alb showed a moderate association with E_NAG in diabetics but not in controls (Figure 3A). It was strongly associated only with those who had FPG ≥110 mg/dL (Figure 3B). E_alb showed a strong association with E_β2M in diabetics (Figure 3C) and those with FPG ≥110 mg/dL (Figure 3D). The associations of E_alb with E_β2M in controls and those with FPG <110 mg/dL were insignificant (Figure 3C,D).

Given the particularly strong correlation strengths of eGFR versus E_NAG (r = −0.467) in the correlation analysis (Table 5), scatterplots and regression analysis were used to further examines the correlations of eGFR with E_NAG/C_cr in subgroups (Figure 4).

eGFR was inversely associated with E_alb only in women (Figure 4A) and those with hypertension (Figure 4B). It showed also inverse association with E_NAG in women only (Figure 4C). In those with hypertension eGFR showed a more closely inverse association with E_NAG, compared to the normotensive group.

2.5. The Indirect Effects of Cd on E_alb and eGFR

Because Cd explain minimally the variations in E_alb and eGFR (Table 4 and Table 5), a simple mediation model analysis was conducted to determine whether Cd could indirectly influence E_alb and eGFR (Figure 5).

In Figure 5, E_NAG was depicted as the mediator of Cd effect on E_alb (model A) and eGFR (model B). In both models A and B, the Sobel test of the direct effects of Cd were insignificant, reflected by β values and p-values for the c′ parameter. The significant indirect effect (a*b) inferred that E_NAG mediated fully the effects of Cd on E_alb and eGFR (Figure 4A,B).

3. Discussion

The present Thai cohort with a modest sample size (65 DM and 72 CTRL) has provided several insights into the effects of low-dose Cd in combination with a high fasting plasma glucose (one form of metabolic stress). Results of logistic regression modeling and covariance analyses of the two major outcomes; albuminuria and a reduced eGFR have revealed the dominant tubular basis of the nephrotoxicity from a low-dose Cd.

3.1. Effects of Cd on FPG and Albuminuria

The levels of Cd exposure experienced by cohort participants were insufficient to induce DM, evident from an insignificant correlation between E_Cd and FPG (r = 0.166, p = 0.053) (Table 5), and the E_Cd geometric mean below 1 µg/g cr. A study from Korea reported an association between increased risk of DM with E_Cd 2-3 µg/g cr, where a significant correlation was found between E_Cd and FPG [15]. In a study on U.S. population, E_Cd of 1–2 μg/g cr was associated with 24-48% increases in risk of prediabetes and diabetes adjusting for potential confounders [16]. Furthermore, macroalbuminuria was present only in the DM group although the DM and CTRL groups both had E_Cd in the same range (Table 1, Figure 1). Thus, the diabetogenic effects of Cd at the exposure levels in this cohort participants appeared to be minimal.

Notably, microalbuminuria (E_cr-based) was found in both CTRL and DM; 8.3% of CTRL had microalbuminuria, while 35.1% of the <10-yr DM and 44% in the ≥10-yr DM had microalbuminuria (Table 1). These data suggest E_alb was particularly sensitive to Cd, consistent with the studies from Spain and China, where the increment of albuminuria risk was observed at E_Cd below 0.5 µg/g cr [18,19]. Compared to the ACR-based, the % microalbuminuria (C_cr-based) were higher in the CTR (12.9), the <10-yr DM (33.3), and the ≥10-yr DM (52.0) groups, indicating the tendency to underestimate the nephrotoxicity of Cd when E_alb was adjusted to E_cr, illustrated in SM (Tables S1-S4).

3.2. Albuminuria and Reduced eGFR: The Tubular Rule

In an analysis of risk factors for albuminuria (Table 2), diabetes, reduced eGFR and hypertension independently increased the risk for albuminuria. Age, BMI, E_Cd/C_cr, smoking and gender had little effects on the prevalence odds for albuminuria in the present cohort. An impaired tubular function, indicated by E_β2M/C_cr ≥ 3 µg/L filtrate and age were independent risk factors for reduced eGFR. Using Ecr-based data, the relationship between a reduced eGFR and risk of albuminuria was obscured as was the relationship between tubular dysfunction and a reduced eGFR (Tables S1 and S2).

Covariance analyses (Table 3 and Table 4) provide further evidence for the tubular roles in albuminuria onset and eGFR reduction. In CTRL, FrTD_β2M accounted for the largest fraction of the variation in E_alb (η² = 0.228), while E_NAG explained the largest fraction of the eGFR variation (η² = 0.301). In DM, gender contributed the most to the variation in E_alb (η² = 0.176) followed by E_NAG (η² = 0.162), hypertension (η² = 0.146), smoking (η² = 0.107), and BMI (η² = 0.097). E_NAG explained the largest fraction of eGFR variation in DM (η² = 0.199), followed by E_β2M (η² = 0.183) and age (η² = 0.140).

The above findings are in line with the SPRINT Trial, where clinical and demographic characteristics were found to be associated with unique profiles of tubular damage, which indicated under-recognized patterns of kidney tubule disease among individuals with reduced eGFR [29].

3.3. Mediation Analysis for the Indirect Effects of Cd

As shown in Table 5, E_alb correlated more strongly with E_NAG than with FrTD_β2M, FPG and E_β2M, while a correlation between E_alb and E_Cd was insignificant. In subgroup analysis (Figure 3), the E_alb/E_NAG association was present only in DM and those with FPG ≥ 110 mg/dL. Of relevance, a cross-sectional study from Japan (DM = 245, CTRL= 39) showed that E_alb/E_cr and E_NAG/E_cr were associated with glycemia risk index, and the elevation of E_NAG/E_cr occurred before the onset of albuminuria.

As also shown in Table 5, eGFR inversely correlated with E_NAG (r = −0.467), and its inverse association with E_NAG was found in women together with an inverse association of eGFR with E_alb (Figure 5). The small number of men (n = 29) compared to women (n = 106) was a likely explanation for the absence of eGFR/E_NAG and eGFR/E_alb associations among men. A further research study with enough men is required. By using mediation analysis (Figure 5), Cd indirectly influenced E_alb and eGFR, while the tubular damage, assessed with E_NAG, was identified as the full mediator of Cd effects on these parameters.

In summary, covariance analysis of albuminuria and a reduced eGFR suggests that the contribution of Cd exposure to albuminuria and eGFR reductions both were minimal (Table 3 and Table 4). Consequently, mediation analysis was used to investigate the indirect effects of Cd (Figure 5). Results of the mediation analysis inferred that tubular injury (E_NAG) could mediate the Cd effects on both albuminuria and a falling eGFR. These findings underscore the essential tubular role in maintaining plasma glucose. and the injury to tubular cells from any causes can potentially influence risk of hyperglycemia. The kidney is known for its producing and releasing glucose into the circulation as well as its filtration and reabsorption of glucose, mediated primarily by SGLT2 [7,8,9,10]. In clinical trial, reduction of glucose reuptake by SGLT2 inhibitors attenuated the loss of eGFR [11].

4. Materials and Methods

4.1. Participants

Individuals with diabetes (n = 65) and non-diabetic controls (n = 72) in the present study were chosen from a pre-existing cohort of 88 persons with diagnosed diabetes and 88 without diabetes [30]. They met the enrollment criteria, which included living at current addresses for at least 40 years and attending annual health checkups at the Pak Poon municipality health center, Nakhon Si Thammarat Province, Thailand. Exclusion criteria were non-resident status, pregnancy, breast feeding and those with records of ill-health status such as heart disease, stroke, and cancer.

Participants gave written informed consent and they received study objectives, study protocols, potential risks and benefits. Sociodemographic information, educational attainment and occupation were collected using structured interview questionnaires as were health status, family history of diabetes, use of dietary supplements, alcohol consumption, and smoking status [31].

4.2. Collection, Storage and Compositional Analysis of Blood and Urine Samples

Whole blood and urine samples were collected in the morning after an overnight fast at the Pak Poon health center. Urine samples were collected in metal-free polypropylene collection cups. For the glucose assay, blood samples were collected in heparinized tubes, using fluoride as an inhibitor of glycolysis. For analysis of blood metals, ethylene diamine tetra acetic acid was an anticoagulant. All test tubes, bottles, and pipettes used in metal analysis were acid-washed and rinsed thoroughly with deionized water.

Blood and urine samples were kept on ice during transportation to the laboratory, where sample aliquots were prepared. The pH of urine aliquots was adjusted to > 6 using 1M NaOH to prevent the degradation of β₂M in acidic conditions. Aliquots of urine, whole blood, serum, and plasma were stored at −80°C for later analysis.

Graphite furnace atomic absorption spectrometry (GBC Scientific Equipment, Hampshire, IL, USA) was employed to determine Cd concentrations in urine and whole blood samples. To prepare samples for AAS, blood samples were deproteinized with 5% HNO₃ as previously described [32]. For the instrumental calibration, the certified reference material of nine elements (Centipur®, Merck KGaA, Darmstadt, Germany) was used. Duplicate samples were analyzed for consistency checks. Blank samples were included for contamination monitoring. Reference urine and whole blood metal control levels 1, 2, and 3 (Lyphocheck, Bio-Rad, Hercules, CA, USA) were used for accuracy and precision assurance and quality control purposes.

The limit of detection (LOD) for blood Pb was 3 µg/dL and 0.1 µg/L for blood Cd. The LOD of urinary Cd was 0.1 µg/L. For urine and blood samples containing Cd and Pb concentrations below the LOD figures, the concentrations assigned were the LOD values divided by the square root of 2 [33].

The urinary NAG assay was based on colorimetry, using 4-nitrophenyl N-acetyl-β-D-glucosaminide as a substrate (Merck KGaA, Darmstadt, Germany). Urinary and serum concentrations of β₂M were determined using the human beta-2 microglobulin/β₂M ELISA pair set (Sino Biological Inc., Wayne, PA, USA). The lower limit of β₂M detection was 3.13 pg/mL. The oxidase–peroxidase method (Glu Colorimetric Assay Kit, Elabscience, Catalog No: E-BC-K234-M, Houston, TX, USA) was used to determine plasma glucose concentrations [34]. Jaffe’s alkaline picrate method was used to determine urinary and plasma creatinine concentrations [35]. All assay gave the coefficients of variation (CV) within acceptable clinical chemistry standards.

4.3. Correction for Differences in Urine Dilution

Because urine samples were collected at a single time point (voided urine), a correction for differences among people in urine volume (dilution) was undertaken. Accordingly, E_alb, E_NAG, E_β2M and E_Cd were normalized to creatinine clearance (C_cr), using the equation; E_x/C_cr = [x]_u[cr]_p/[cr]_u, where x = alb, NAG, β₂M or Cd; [x]_u = urine concentration of x (mass/volume); [cr]_p = plasma creatinine concentration (mg/dL); and [cr]_u = urine creatinine concentration (mg/dL). E_x/C_cr was expressed as an amount of x excreted per volume of the glomerular filtrate [36]. This C_cr-normalization is unaffected by variation in muscle mass, while correcting for both dilution and functioning nephrons.

E_alb, E_NAG, E_β2M and E_Cd were normalized to creatinine excretion (E_cr), using the equation; E_x/E_cr = [x]_u/[cr]_u, where x = alb, NAG, β₂M or Cd; [x]_u = urine concentration of x (mass/volume) and [cr]_u = urine creatinine concentration (mg/dL). E_x/E_cr was expressed as an amount of x excreted per g of creatinine. Results of logistic regression and covariance analysis conducted with Ecr-based data are provided in SM (Tables S1–S4).

E_cr-normalization corrects for difference in urine dilution only and it is affected by a large variation in muscle mass, especially between men and women. E_cr-adjustment can create non-differential errors/imprecisions that can obscure or even nullify the dose-response relationships [37]. Previously, the risk of abnormal E_β2M values was not significantly related to E_Cd when E_Cd and E_β2M were adjusted to E_cr [38].

4.4. Computaion for eGFR and Fractional Tubular Degrdation of β₂M

Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations were used to compute eGFR [39]. CKD stages 1, 2, 3, 4, and 5 corresponded to eGFR of 90–119, 60–89, 30–59, 15–29, and <15 mL/min/1.73 m², respectively.

The fractional tubular degradation of filtered β₂M (FrTD_β2M) (equation 1) was computed using rate of glomerular filtration of β₂M (F_β2M), mg/d (equation 2) and amount of β₂M undergoing tubular degradation per volume of glomerular filtrate (TD_β2M/C_cr), mg/L filtrate) (equation 3) [40].

Equation 1: FrTD_β2M = (TD_β2M/C_cr)/[β₂M]_s

Equation 2: F_β2M = eGFR[β₂M]_s

Equation 3: TD_β2M/C_cr = [β₂M]_s – E_β2M/C_cr

4.5. Analysis for the Indirect Effects of Cd

The Baron and Kenny method was used to examine the indirect and/or direct effects Cd on E_alb and eGFR [41,42,43]. The mediation model with one mediator is depicted in Scheme 1.

The statistical parameters a, b, and c′ are β coefficients describing the effect of IV on M, the effect of M on DV and the direct effect of IV on DV, respectively. The product a*b indicates the indirect effect of IV on DV, mediated by M. The Sobel test is employed to define a statistical significance level of the indirect effect (a*b).

4.6. Statistical Analysis

Data were analyzed with IBM SPSS Statistics 21 (IBM Inc., New York, NY, USA). The Kruskal–Wallis’s test was employed to assess the variation in any continuous variable across three study groups; CTRL, <10-yr DM and ≥ 10-yr DM. The Pearson chi-squared test assessed differences in percentages across the three study groups. The one-sample Kolmogorov–Smirnov test assessed deviation from a normal distribution of any continuous variable. Logarithmic transformation was applied to E_Cd/C_cr, E_alb/C_cr, E_β2M/C_cr and E_NAG/C_cr that showed rightward skewing before they were subjected to parametric statistics analyses, scatterplots, and linear regressions.

The POR values for albuminuria and a reduced eGFR were obtained using logistic regression modeling, which adjusted for potential confounders (Table 2). The univariate of analysis of covariance was undertaken to identify the contributors to the variation in E_alb and eGFR (Table 3 and Table 4). Spearman’s rank correlation analysis was used to assess bivariate relationships of nine variables: E_alb/C_cr, age, BMI, eGFR, FPG, E_Cd/C_cr, E_NAG/C_cr, E_β2M/C_cr and FrTD_β2M (Table 5).

5. Conclusions

Exposure to low-level Cd, insufficient to induce abnormally high plasma glucose concentrations, is not associated with an increased risk of albuminuria or reduced eGFR. However, in the presence of diabetes and/or hyperglycemia, exposure to low-level Cd could enhance albumin excretion rates, while aggravating eGFR reduction through inducing tubular cell damage and functional impairment.