Cadmium-Induced Tubular Dysfunction in Type 2 Diabetes: A Population-Based Cross-Sectional Study

The global prevalence of diabetes, and its major complication, diabetic nephropathy, have reached epidemic proportions. The toxic metal cadmium (Cd) also induces nephropathy, indicated by a sustained reduction in the estimated glomerular filtration rate (eGFR) and the excretion of β2-microglobulin (β2M) above 300 µg/day, which reflects kidney tubular dysfunction. However, little is known about the nephrotoxicity of Cd in the diabetic population. Here, we compared Cd exposure, eGFR, and tubular dysfunction in both diabetics (n = 81) and non-diabetics (n = 593) who were residents in low- and high-Cd exposure areas of Thailand. We normalized the Cd and β2M excretion rates (ECd and Eβ2M) to creatinine clearance (Ccr) as ECd/Ccr and Eβ2M/Ccr. Tubular dysfunction and a reduced eGFR were, respectively, 8.7-fold (p < 0.001) and 3-fold (p = 0.012) more prevalent in the diabetic than the non-diabetic groups. The doubling of ECd/Ccr increased the prevalence odds ratios for a reduced eGFR and tubular dysfunction by 50% (p < 0.001) and 15% (p = 0.002), respectively. In a regression model analysis of diabetics from the low-exposure locality, Eβ2M/Ccr was associated with ECd/Ccr (β = 0.375, p = 0.001) and obesity (β = 0.273, p = 0.015). In the non-diabetic group, Eβ2M/Ccr was associated with age (β = 0.458, p < 0.001) and ECd/Ccr (β = 0.269, p < 0.001). However, after adjustment for age, and body mass index (BMI), Eβ2M/Ccr was higher in the diabetics than non-diabetics of similar ECd/Ccr ranges. Thus, tubular dysfunction was more severe in diabetics than non-diabetics of similar age, BMI, and Cd body burden.


Introduction
Type 2 diabetes is a metabolic disorder resulting in a rise of fasting plasma glucose ≥ 126 mg/dL. The worldwide prevalence of diabetes, which is often linked to obesity, has now reached epidemic proportions. However, it is increasingly apparent that exposure to various diabetogenic pollutants, such as cadmium (Cd), is an important risk factor [1][2][3][4][5][6]. Strong evidence supporting the diabetogenicity of Cd comes from the Wuhan-Zhuhai prospective cohort study. This study, which measured the fasting plasma glucose levels and urinary Cd over a 3-year period, reported that for each tenfold increase in urinary Cd, the prevalence of prediabetes increased by 42% [4].
Although Cd exposure has only recently been recognized as a risk factor for diabetes [5,6], the increased susceptibility to Cd-induced nephrotoxicity in people with diabetes was first noted in the 1990 Belgian population study (Cadmibel) [7]. Similar observations were then made in studies from Sweden [8,9], Australia [10], the U.S. [11], and Korea [12]. The defective tubular reabsorption of proteins, indicated by an increase in the excretion of proteins of low molecular weight, such as β 2 -microglobulin (β 2 M), is the most frequently reported sign of Cd nephrotoxicity [13,14]. In early studies, the

Cohort Participants
Participants came from three population-based studies undertaken in a Cd pollution area in the Mae Sot District, Tak Province (n = 211), two low-Cd-exposure locations in Bangkok (n = 322), and the Pakpoon municipality of Nakhon Si Thammarat Province (n = 141). The Institutional Ethical Committees of Chulalongkorn University, Chiang Mai University, and the Mae Sot Hospital approved the study protocol for the Mae Sot and Bangkok groups [30]. The Office of the Human Research Ethics Committee of Walailak University in Thailand approved the study protocol for the Pakpoon group [31]. All participants gave informed consent, and all had resided at their current addresses for at least 30 years. Exclusion criteria included pregnancy, breastfeeding, a history of metal work, and a hospital record or physician's diagnosis of an advanced chronic disease.
Smoking, diabetes, hypertension, regular use of medications, educational level, occupation, and family health history were ascertained by questionnaire. Prediabetes and diabetes were indicated by fasting plasma glucose levels ≥110 and ≥126 mg/dL, respectively (https://www.cdc.gov/diabetes/basics/getting-tested.html) (accessed on 17 February 2023) or a physician's prescription of anti-diabetic medications. Hypertension was defined as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, a physician's diagnosis, or prescription of anti-hypertensive medications.
As workplace exposure was an exclusion criterion, participants presumably acquired Cd from the diet and/or smoking. Based on the measured levels of Cd in duplicate diets [32] and a nationwide survey of Cd levels in soils and food crops [33], environmental exposure to Cd in Bangkok and Nakhon Si Thammarat were low, and the Mae Sot district was high. From a previous study conducted in the Mae Sot district, the Cd content of the paddy soil samples exceeded the standard of 0.15 mg/kg, and rice samples collected from households contained 4 times the amount of the permissible Cd level of 0.1 mg/kg [34]. A health survey reported that the prevalence of CKD in the Mae Sot was 16.1%, while the prevalence of tubular proteinuria was 36.1% [35].

Urine and Blood Sampling and Analysis
Second morning urine samples were collected after an overnight fast. Samples of whole blood were obtained within 3 h of urine sampling. Aliquots of urine, whole blood, and plasma were stored at −20 • C or −80 • C for later analysis. The assay for the urine and plasma concentrations of creatinine ([cr] u and [cr] p ) was based on the Jaffe reaction. The assay of β 2 M in the urine ([β 2 M] u ) was based on the latex immunoagglutination method (LX test, Eiken 2MGII; Eiken and Shionogi Co., Tokyo, Japan).
For the Bangkok group, the urinary concentration of Cd ([Cd] u ) was determined by inductively coupled plasma mass spectrometry (ICP/MS, Agilent 7500, Agilent Technologies, Santa Clara, CA, USA). Multi-element standards (EM Science, EM Industries, Inc., Newark, NJ, USA) were used to calibrate the Cd analyses. The accuracy and precision of those analyses were ascertained with reference urine (Lyphochek ® , Bio-Rad, Sydney, Australia). The [Cd] u assigned to samples with Cd below the detection limit was 0.05 µg/L divided by the square root of 2 [36].
For the Pakpoon group, the [Cd]u was determined with the GBC System 5000 graphite furnace atomic absorption spectrophotometer (AAS) (GBC Scientific Equipment, Hampshire, IL, USA). Multielement standards were used to calibrate the metal analysis (Merck KGaA, Darmstadt, Germany). Reference urine levels 1, 2, and 3 (Lyphocheck, Bio-Rad, Hercules, CA, USA) were used for quality control, analytical accuracy, and precision assurance. When a [Cd] u level was less than its detection limit, the concentration assigned was 0.1 µg/L divided by the square root of 2 [36].
For the Mae Sot group, the [Cd]u was determined with AAS (Shimadzu Model AA-6300, Kyoto, Japan). Urine standard reference material No. 2670 (National Institute of Standards, Washington, DC, USA) was used for quality assurance and control purposes. The limit of detection of the [Cd] u was 0.06 µg/L. None of the urine samples contained a [Cd] u below the detection limit.
The comparability of the [Cd] u was ascertained by simultaneous quantification of Cd in the reference urine samples where the coefficient of variation was within acceptable clinical chemistry standards.

Estimated Glomerular Filtration Rate
The GFR is the product of the nephron number and the mean single nephron GFR, and, in theory, the GFR is indicative of nephron function [28,37]. In practice, the GFR is estimated from established CKD-EPI equations and reported as eGFR [28]. The

Statistical Analysis
The data were analyzed with IBM SPSS Statistics 21 (IBM Inc., New York, NY, USA). We used the Mann-Whitney U-test to assess differences in the means between the two groups. The Pearson chi-squared test was used to assess differences in the percentages. To identify the departures of continuous variables from a normal distribution, the onesample Kolmogorov-Smirnov test was used, and logarithmic transformation was applied to variables that showed rightward skewing before they were subjected to parametric statistical analysis. Logistic regression analysis was used to determine the prevalence odds ratio (POR) for tubular proteinuria and reduced eGFR. Tubular proteinuria was defined as E β2M /C cr ≥ 300 µg/L of filtrate. Reduced eGFR was assigned when eGFR ≤ 60 mL/min/1.73 m 2 . Univariate/covariance analyses with Bonferroni correction in multiple comparisons were used to obtain the mean E β2M /C cr adjusted for age and BMI, and interactions among groups of diabetics and non-diabetics stratified by three ranges of E Cd /C cr, . For all tests, p-values ≤ 0.05 were considered to indicate statistical significance.

Cohort Composition and Characteristics
The Thai cohort of 674 participants consisted of 463 drawn from two low-exposure localities, and 211 from an area where environmental Cd pollution is endemic (Table 1). Females constituted 57.6% of the cohort, and they formed 74.1% of the diabetic group (n = 81). Seventy subjects with diagnosed diabetes (86.4%) were residents of the lowexposure location, Pakpoon, and 11 diabetic cases were from a high-exposure area. The diabetic group was older (58.5 years) than the non-diabetic group (45.6 years), and the overall mean age was 47.2 years. The percentage (%) of smoking was lower in diabetics Toxics 2023, 11, 390 5 of 14 (14.9%) than in non-diabetics (34.2%). The % of obese participants was higher in the diabetics than in the non-diabetics (14.8% vs. 2.4%).
The diabetic group had a lower mean eGFR and a higher % of reduced eGFR, defined as eGFR ≤ 60 mL/min/1.73 m 2 , compared to non-diabetics (22.2% vs. 9.9%). The mean E Cd /C cr and mean E Cd /E cr in diabetics were all lower than in non-diabetics. Conversely, the mean E β2M /C cr and mean E β2M /E cr in diabetics were all higher than in non-diabetics. Nearly half (49.4%) of the diabetic group and 14.8% of the non-diabetic group had severe tubular proteinuria (E β2M /C cr values ≥ 1000 µg/L filtrate).

Predictors of Tubular Proteinuria and Reduced eGFR
We employed logistic regression analysis to screen factors that may increase the risk of tubular proteinuria and/or reduced eGFR. In this analysis, the independent variables were age, BMI, log 2 [(E Cd /C cr ) × 10 5 ], hypertension, smoking, gender, and diabetes ( Table 2). Abbreviations: POR, prevalence odds ratio; CI, confidence interval. Coding: female = 1, male = 2, normotension = 1, hypertension = 2, non-smoker = 1, smoker = 2. a Tubular dysfunction was defined as (E β2M /C cr ) × 100 ≥ 300 µg/L filtrate; b reduced eGFR was defined as estimated GFR ≤ 60 mL/min/1.73 m 2 . Data were generated from logistic regression analyses relating POR to tubular proteinuria and reduced eGFR to a set of seven independent variables (first column). First number and number in parenthesis in second column correspond to number of participants in analyses of tubular dysfunction and reduced eGFR, respectively. For all tests, p-values ≤ 0.05 indicate a statistically significant association of POR with a given independent variable.
BMI, hypertension, smoking, and gender did not show significant associations with tubular dysfunction or a reduced eGFR. Three other independent variables, namely, age, diabetes, and E Cd /C cr , were all associated with both tubular proteinuria and reduced eGFR.
Tubular dysfunction and reduced eGFR were more prevalent in the diabetic than the non-diabetic groups by 8.7-fold (p < 0.001) and by 3-fold (p = 0.012), respectively. For a one-year increase in age, the POR for tubular dysfunction and reduced eGFR rose by 10% (p < 0.001) and 15% (p < 0.001), respectively. The doubling of E Cd /C cr was associated with a 15% increase in the POR for tubular dysfunction (p = 0.002) and a 50% increase in the POR for reduced eGFR (p < 0.001).

Effects of Cadmium and Diabetes on β 2 M Excretion
We further evaluated the effects of diabetes and Cd exposure on E β2M by scatterplots and covariance analyses (Figure 1).
A direct relationship was seen between E β2M /C cr and E Cd /C cr in the non-diabetic group (R 2 0.136, p < 0.001) (Figure 1a). After controlling for interactions and differences in age and BMI, the mean log[(E β2M /C cr ) × 10 4 ] was, respectively, the highest, in the middle, and lowest in non-diabetics who had high, moderate, and low E Cd /C cr ranges (F = 24.08, p < 0.001) (Figure 1b).

Effects of Cadmium and Diabetes on β2M Excretion
We further evaluated the effects of diabetes and Cd exposure on Eβ2M by scatterplots and covariance analyses (Figure 1). A direct relationship was seen between Eβ2M/Ccr and ECd/Ccr in the non-diabetic group (R 2 0.136, p < 0.001) (Figure 1a). After controlling for interactions and differences in age and BMI, the mean log[(Eβ2M/Ccr) × 10 4 ] was, respectively, the highest, in the middle, and lowest in non-diabetics who had high, moderate, and low ECd/Ccr ranges (F = 24.08, p < 0.001) (Figure 1b).
In the analysis, including all 81 diabetics (Figure 1a), the relationship between Eβ2M/Ccr and ECd/Ccr did not reach a statistically significant level (p = 0.053), and the variation in the mean log[(Eβ2M/Ccr) × 10 4 ] across the three ECd/Ccr ranges was insignificant (F = 0.204, p = 0.816) (Figure 1b). However, the relationship between Eβ2M and ECd was significant when 11 diabetic cases from a high-Cd exposure area were excluded (R 2 0.127, p = In the analysis, including all 81 diabetics (Figure 1a), the relationship between E β2M /C cr and E Cd /C cr did not reach a statistically significant level (p = 0.053), and the variation in the mean log[(E β2M /C cr ) × 10 4 ] across the three E Cd /C cr ranges was insignificant (F = 0.204, p = 0.816) (Figure 1b). However, the relationship between E β2M and E Cd was significant when 11 diabetic cases from a high-Cd exposure area were excluded (R 2 0.127, p = 0.002 (Figure 1c). After adjustment for age and BMI, the mean log[(E β2M /C cr ) × 10 4 ] was higher in diabetic subjects of low and moderate ranges of E Cd /C cr than in non-diabetics of similar E Cd /C cr ranges (Figure 1d). For the highest E Cd /C cr subsets, the mean log[(E β2M /C cr ) × 10 4 ] in the diabetics and non-diabetics was similar.
The results of the regression analyses of β 2 M excretion are provided separately for the diabetics, diabetics from a low-exposure area, and non-diabetics (Table 3). In a model including 70 diabetic cases from the low-exposure locality (Pakpoon), 24.4% of the variation in E β2M /C cr was explained by all six independent variables. E β2M /C cr varied directly with E Cd /C cr (β = 0.375, p = 0.001) and obesity (β = 0.273, p = 0.015). However, when all 81 diabetics were included in the analysis, only 8.9% of the E β2M /C cr variation was explained by age, log 2 [(E Cd /C cr ) × 10 5 ], smoking, obesity, gender, and hypertension. In effect, none of these six variables showed a significant association with E β2M /C cr in the diabetic group.
In the non-diabetic group, the six independent variables explained 38.6% of the E β2M /C cr variability, where E β2M /C cr varied directly with age (β = 0.458, p < 0.001) and with E Cd /C cr (β = 0.269, p < 0.001).

Inverse Association of eGFR and Cadmium
Similarly, we used multiple regression analyses to compare the strength of the association of the eGFR and E Cd /C cr in three subsets (Table 4). In a model including all diabetics, eGFR was inversely associated with age (β = −0.444) and E Cd /C cr (β = −0.244), and these two variables, plus smoking, obesity, gender, and hypertension accounted for 33% of the eGFR variability. These six independent variables explained 29.3% of the total variation in eGFR among diabetics from a low-Cd-exposure area, and only age showed a significant association with eGFR reduction (β = −0.472). An association between the eGFR and E Cd /C cr was insignificant (β = −0.145, p = 0.167).
In the non-diabetic group, age, E Cd /C cr , smoking, obesity, gender, and hypertension together accounted for 53.4% of the total eGFR variation. Distinct from the diabetics from a low-exposure area, the eGFR among those without diabetes was inversely associated with age (β = −0.574) and E Cd /C cr (β = −0.263).

Inverse Association of β2M Excretion and eGFR
To assess the association of E β2M /C cr with the eGFR, we employed scatterplots and covariance analyses, where differences in age and BMI were adjusted together with interactions ( Figure 2). cate the strength of the association of the eGFR and independent variables. Adjusted R 2 indicates the proportion of the variation in eGFR attributable to all six independent variables.
In a model including all diabetics, eGFR was inversely associated with age (β = −0.444) and ECd/Ccr (β = −0.244), and these two variables, plus smoking, obesity, gender, and hypertension accounted for 33% of the eGFR variability. These six independent variables explained 29.3% of the total variation in eGFR among diabetics from a low-Cd-exposure area, and only age showed a significant association with eGFR reduction (β = −0.472). An association between the eGFR and ECd/Ccr was insignificant (β = −0.145, p = 0.167).
In the non-diabetic group, age, ECd/Ccr, smoking, obesity, gender, and hypertension together accounted for 53.4% of the total eGFR variation. Distinct from the diabetics from a low-exposure area, the eGFR among those without diabetes was inversely associated with age (β = −0.574) and ECd/Ccr (β = −0.263).

Inverse Association of β2M Excretion and eGFR
To assess the association of Eβ2M/Ccr with the eGFR, we employed scatterplots and covariance analyses, where differences in age and BMI were adjusted together with interactions ( Figure 2). In the diabetic group, there was a strong inverse relationship between Eβ2M/Ccr and the eGFR (R 2 0.412, p < 0.001) (Figure 2a), and the eGFR explained 33.4% of the Eβ2M/Ccr variation across the three eGFR ranges (Figure 2b). A large proportion of the Eβ2M/Ccr variation was explained by a single variable, eGFR.  In the diabetic group, there was a strong inverse relationship between E β2M /C cr and the eGFR (R 2 0.412, p < 0.001) (Figure 2a), and the eGFR explained 33.4% of the E β2M /C cr variation across the three eGFR ranges (Figure 2b). A large proportion of the E β2M /C cr variation was explained by a single variable, eGFR.
Similarly, an inverse relationship was seen between E β2M /C cr and E Cd /C cr in the non-diabetic group (R 2 0.400, p < 0.001) (Figure 2c). The mean log[(E β2M /C cr ) × 10 4 ] was the highest, in the middle, and lowest in those with an eGFR ≤ 60, 61-90, and >90 mL/min/1.73m 2 , respectively (Figure 2d). The eGFR explained only 10.3% of the E β2M /C cr variation across these ranges of the eGFR (F = 24.48, p < 0.001). This variation in E β2M /C cr attributable to the eGFR was smaller compared to the diabetic group (Figure 2b).

β2M Excretion as a Function of GFR and Kidney Cadmium Burden
Because the GFR showed a strong influence on β 2 M excretion (Figure 2), we next compared E β2M /C cr in subsets with a normal or reduced eGFR across three E Cd /C cr ranges (<1, 1-4.99 and ≥5 ng/L filtrate). The results of these analyses are shown in Figure 3.   Among 81 diabetics, 63 had normal eGFR, while 18 (22.2%) had reduced eGFR (Figure 3a,b). A direct relationship between log[(E β2M /C cr ) × 10 4 ] and log[(E Cd /C cr ) × 10 5 ] was seen only in the reduced eGFR group (R 2 0.493, p < 0.001). In covariance analysis (Figure 2b), the mean log[(E β2M /Ccr) × 10 4 ] tended to be higher in the low eGFR subsets of all three E Cd /C cr ranges. The overall mean log[(E β2M /C cr ) × 10 4 ] was higher in the low eGFR than that of the high eGFR group (p < 0.001).
Among the 454 non-diabetics, 395 had a normal eGFR, while 59 (13%) had a reduced eGFR (Figure 3c,d). Log[(E β2M /C cr ) × 10 4 ] rose with log[(E Cd /C cr ) × 10 5 ] in both the low and normal eGFR groups. Like the diabetics, the mean log[(E β2M /Ccr) × 10 4 ] tended to be higher in the low eGFR subsets of all three E Cd /C cr ranges. The overall mean log[(E β2M /C cr ) × 10 4 ] was higher in the reduced eGFR than in the normal eGFR group (p < 0.001).

Discussion
In the present study, we compared the severity of Cd-induced nephropathy in diabetics and non-diabetics living in low-and high-Cd-exposure areas of Thailand. The prevalence of a reduced eGFR (below 60 mL/min/1.73 m 2 ) in our cohort was 11.4% which is higher than that reported in studies from Spain (7%) [38] and Taiwan (6.3%) [39], but in line with the global prevalence of CKD, which varies between 8% and 16% [40]. Approximately, one in five cohort participants (27.6%) had tubular dysfunction, based on the conventional E β2M /E cr ≥ 300 µg/g creatinine criterion (Table 1). We found that BMI, hypertension, smoking, and gender were independent variables that showed no significant associations with the risk for tubular proteinuria or a reduced eGFR, but age, diabetes, and measured long-term Cd exposure (body burden) did (Table 2).
For every one-year increase in age, the risks of tubular dysfunction and a reduced eGFR increased by 10% and 15%, respectively. The doubling of the Cd body burden increased the risk of tubular dysfunction by 15% while raising the risk of reduced eGFR by 50%. Thus, Cd had a particularly strong effect on the GFR in this population. This is consistent with the results of many other studies, which have linked an elevated risk of a reduced eGFR to environmental exposure to Cd, including studies from the U.S. [11,[17][18][19], Thailand [20] Guatemala [21], Myanmar [22], Taiwan [23], and Korea [24,25].
The influence of both Cd body burden and the eGFR on E β2M /C cr were evident when participants were stratified by the eGFR and ranges of E Cd /C cr , a measure of Cd body burden. In the diabetic group, 33.4% of the variation in E β2M /C cr was associated with the eGFR (Figure 2b). This is a very large variation in the E β2M /C cr that was due to a single variable, eGFR. In comparison, the eGFR explained only 10.3% of the E β2M /C cr variation in the non-diabetic group (Figure 2b). After adjustment for age and BMI, the mean values of E β2M /C cr in the diabetic and non-diabetic subsets were found to be higher in those with reduced eGFR compared to the normal eGFR subsets of similar Cd body burden (Figure 3a,b).
Evidence for increased susceptibility to Cd-induced tubulopathy among the diabetics comes from covariate analysis, where a relationship between E β2M /C cr and E Cd /C cr was seen only in those from a low-Cd-exposure region (Figure 1, Table 3). These findings are in line with the published reports showing the high susceptibility to the nephrotoxicity of Cd among people with diabetes, as discussed below.
The Cadmibel study found that diabetics were more susceptible than non-diabetics to Cd-induced nephrotoxicity [7]. A similar observation followed in studies conducted in Sweden [8,9], the Australian Torres Strait, [10], the U.S. [11], and Korea [12]. Experimental studies have shown that nephropathy due to diabetes and Cd are magnified when both the metal and the disease are present. The injection of Cd-MT complexes into obese diabetic mice and non-obese littermates resulted in increased urinary excretion of proteins and calcium in both groups [41]. However, in the diabetic mice, the dose of Cd-MT required to induce proteinuria and calciuria was one-fourth of that required in the controls. Cd-MT induced glycosuria in both groups. Chinese hamsters with hereditary diabetes are also highly susceptible to Cd-induced nephrotoxicity [42]. In recent histopathological studies, kidney tubular degeneration and fibrosis due to Cd were more pronounced in diabetic than non-diabetic rats [43,44].
Increases in the risks of prediabetes and diabetes among U.S. adults have been associated with E Cd /E cr of 1-2 µg/g creatinine [1,2]. In a community-based study in Dallas, Texas, an elevated risk of diabetes was linked to environmental Cd exposure [3]. In a metaanalysis of pooled data from 42 studies, the risks of prediabetes and diabetes increased linearly with blood and urinary Cd; the risk of prediabetes reached a plateau at an E Cd /E cr rate of 2 µg/g creatinine, and the diabetic risk rose as blood Cd reached 1 µg/L [5,6]. Of note, these urinary Cd and blood Cd levels have also been in the range associated with a reduced eGFR in studies conducted in many countries listed previously. In a Chinese population study, dietary Cd exposure estimates of 23.2, 29.6, and 36.9 µg/d were associated with 1.73-, 2.93-, and 4.05-fold increments in the prevalence of CKD, compared to a 16.7 µg/d intake level [45]. A diet high in rice, pork, and vegetables was associated with a 4.56-fold increase in the prevalence of CKD [45].
Smoking has been shown to promote both the onset and progression of CKD [46,47]. In a meta-analysis of data from 104 studies, an increase in the odds of CKD of 18% was seen among current and former smokers compared to those who never smoked [46]. A Singaporean prospective cohort (n = 63,257, 30.6% were smokers) has implicated smoking in the progression of CKD [47]. With adjustment for confounders, smoking increased the risk of end-stage kidney disease by 29% compared to non-smokers, while the risk of kidney failure diminished after quitting smoking for more than 10 years [47]. A strong dose-dependent association was seen between the number of years of smoking and kidney failure [47].
In the present study, the risk of CKD (reduced eGFR) was not associated with smoking; instead, it was associated with an indicator of cumulative Cd exposure from all sources. Per the doubling of Cd body burden, there was a 50% increase in the prevalence odds of CKD (Table 2). Smoking is a significant source of Cd exposure, given that cigarette smoke contains Cd in volatile metallic and oxide (CdO) forms, which have transmission rates 5 to 10 times higher than those that enter through the gut [48]. Cd exposure through smoking has been found to increase the risk of diabetic nephropathy in a Dutch cross-sectional study, including 231 patients with type 2 diabetes, where active smokers were found to have significantly higher blood Cd compared to never smokers and former smokers [49]. Data also demonstrated that smoke-derived Cd mediated this nephrotoxicity [49]. In a six-year median follow-up of these 231 diabetic patients, both Cd and active smoking were associated with progressive eGFR reduction [50]. Collectively, findings from the Dutch cohorts support the premise that exposure to even low levels of environmental Cd promotes the development and progression of diabetic kidney disease. This lends support to our observation that people with diabetes are more susceptible to Cd-induced tubulopathy than non-diabetics.

Conclusions
This study shows that tubular dysfunction and a reduced eGFR are more severe and more prevalent in diabetics than non-diabetics of similar age, BMI, and Cd body burden.