The Validity of Benchmark Dose Limit Analysis for Estimating Permissible Accumulation of Cadmium

Cadmium (Cd) is a toxic metal pollutant that accumulates, especially in the proximal tubular epithelial cells of kidneys, where it causes tubular cell injury, cell death and a reduction in glomerular filtration rate (GFR). Diet is the main Cd exposure source in non-occupationally exposed and non-smoking populations. The present study aimed to evaluate the reliability of a tolerable Cd intake of 0.83 μg/kg body weight/day, and its corresponding toxicity threshold level of 5.24 μg/g creatinine. The PROAST software was used to calculate the lower 95% confidence bound of the benchmark dose (BMDL) values of Cd excretion (ECd) associated with injury to kidney tubular cells, a defective tubular reabsorption of filtered proteins, and a reduction in the estimated GFR (eGFR). Data were from 289 males and 445 females, mean age of 48.1 years of which 42.8% were smokers, while 31.7% had hypertension, and 9% had chronic kidney disease (CKD). The BMDL value of ECd associated with kidney tubular cell injury was 0.67 ng/L of filtrate in both men and women. Therefore, an environmental Cd exposure producing ECd of 0.67 ng/L filtrate could be considered as Cd accumulation levels below which renal effects are likely to be negligible. A reduction in eGFR and CKD may follow when ECd rises from 0.67 to 1 ng/L of filtrate. These adverse health effects occur at the body burdens lower than those associated with ECd of 5.24 µg/g creatinine, thereby arguing that current health-guiding values do not provide a sufficient health protection.


Introduction
Cadmium (Cd) is an environmental toxicant of worldwide public health significance because numerous population-based studies suggest that exposure to Cd even at low levels adversely impacts the functions of most organs of the body [1]. Volcanic emissions, biomass and fossil fuel combustion, and cigarette smoke are sources of environmental Cd pollution [2][3][4][5][6]. Cd in cigarette smoke as a volatile metallic form and oxide (CdO) has a particularly high transmission rate [7]. The utility of Cd in many industrial processes, and the use of phosphate fertilizers contaminated with Cd cause also a widespread dispersion of this toxic metal in the environment and subsequently the food chains [8][9][10][11][12].
Foods that are frequently consumed in large quantities such as rice, potatoes, wheat, leafy salad vegetables, and other cereal crops form the most significant dietary sources of Cd [13][14][15]. Seafood (shellfish), molluscs and crustaceans are additional dietary Cd sources [16,17].
Cd accumulates most extensively in the proximal tubular epithelial cells of the kidneys, and the kidney burden of Cd as µg/ g tissue weight increases with age [18,19]. Approximately 0.001-0.005% of Cd in the body is excreted in the urine each day, and the biological half-life of Cd in the kidney cortex is estimated to be 30 years for non-smokers [19,20]. In Japanese residents of a Cd pollution area, the average half-life of the metal among those with a lower body burden (urinary Cd < 5 µg/L) was 23.4 years; in those with a higher body burden (urinary Cd > 5 µg/L), the average half-life was 12.4 years [21,22]. Thus, the lower the body burden, the longer the half-life of Cd.
To safeguard against excessive dietary Cd exposure, health guidance values have been established by the Joint FAO/WHO Expert Committee on Food Additives and Contaminants (JECFA) [23]. Based on a risk assessment model that assumed an increase in the excretion of β 2 -microglobulin (β 2 M, E β2M ) above 300 µg/g creatinine to be the "critical" toxicity endpoint, the tolerable intake level of Cd was set at 0.83 µg/kg body weight/day, and a urinary Cd excretion of 5.24 µg/g creatinine was considered to be the toxicity threshold level [23]. However, E β2M levels of 100-299, 300-999, and ≥1000 µg/g creatinine were found to be associated with 4.7-, 6.2-and 10.5-fold increases in the risk of an estimated GFR (eGFR) ≤ 60 mL/min/1.73 m 2 , commensurate with chronic kidney disease (CKD) [24]. Thus, E β2M above 300 µg/g creatinine does not appear to be an early warning sign of the nephrotoxicity of Cd. The utility of β 2 M excretion as a toxicity criterion to derive Cd health guidance values is questionable.
Presently, the benchmark dose (BMD) has often been used to derive health guidance values instead of the no-observed-adverse-effect level (NOAEL) [25][26][27]. The BMD is a dose level, derived from an estimated dose-response curve, associated with a specified change in response, termed benchmark response (BMR) which can be set at 1%, 5%, and 10% as required. BMD rectifies some of the shortcomings of the NOAEL [28,29]. For a continuous endpoint, the lower 95% confidence bound of BMD, termed BMDL value derived at 5% BMR has been viewed as NOAEL equivalent or the level below which adverse health effect could be negligible [25].
The present study had three aims. The first aim was to evaluate the validity of current kidney toxicity threshold level of E Cd at 5.24 µg/g creatinine. The second aim was to define the BMDL values (NOAEL equivalents) of E Cd from three indicators of renal effects; injury to kidney tubular epithelial cells, a defective tubular reabsorption of β 2 M, and eGFR decline. Kidney tubular cell injury was indicated by an increase in excretion of N-acetyl-β-D-glucosaminidase (NAG, E NAG ). We postulated that the BMDL value of E Cd for the E NAG endpoint was the lowest, given that the E NAG represents an early sign of the cytotoxicity of Cd. The third aim was to demonstrate an impact of methods used to normalize Cd, NAG and β 2 M excretion rates. Accordingly, we assessed the validity of BMDL values of E Cd obtained from normalizing excretion rates of Cd, NAG and β 2 M to creatinine clearance (C cr ) as E x /C cr and to the excretion of creatinine (E cr ) as E x /E cr , where x = Cd, NAG or β 2 M.

Study Subjects
We assembled archived data from 289 men and 445 women who were drawn from Bangkok (a low-exposure area) and from Cd contaminated areas of Mae Sot District (a high-exposure area) in Thailand [30]. The Ethical Committee of Chulalongkorn University and the Mae Sot Hospital Ethical Committee approved the study protocol. At the time of recruitment, all participants had lived at their current addresses for at least 30 years, and all gave informed consent prior to participation. Exclusion criteria were pregnancy, breastfeeding, a history of metalwork, and a hospital record or physician's diagnosis of an advanced chronic disease. Smoking, diabetes, hypertension, regular use of medications, educational level, occupational and family health history were ascertained by a questionnaire. Diabetes was defined as fasting plasma glucose levels ≥ 126 mg/dL or a physician's prescription of anti-diabetic medications. Hypertension was defined as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, and/or a physician's diagnosis and prescription of anti-hypertensive medications.

Blood and Urine Sampling and Analyses
Simultaneous blood and urine sampling is required to normalize E Cd , E β2M and E NAG to C cr Accordingly, second morning urine samples were collected after an overnight fast, and whole blood samples were obtained within 3 h after the urine sampling. Aliquots of urine, whole blood and plasma were stored at −20 • C or −80 • C for later analysis. The assay for urine and plasma concentrations of creatinine ([cr] u and [cr] p ) was based on the Jaffe reaction. The urinary NAG assay was based on colorimetry (NAG test kit, Shionogi Pharmaceuticals, Sapporo, Japan). The urinary β 2 M assay was based on the latex immunoagglutination method (LX test, Eiken 2MGII; Eiken and Shionogi Co., Tokyo, Japan).
For the Bangkok group, [Cd] u was determined by inductively-coupled plasma mass spectrometry (ICP/MS, Agilent 7500, Agilent Technologies, Santa Clara, CA, USA) because it had the high sensitivity required to measure very low Cd concentrations. 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 low limit of detection (LOD) of urine Cd, calculated as 3 times the standard deviation of blank measurements was 0.05 µg/L. The Cd concentration assigned to samples with Cd below the detection limit was the detection limit divided by the square root of 2 [31].
Based on E Cd data reported for Mae Sot residents that could be reliably quantified by atomic absorption spectrophotometry (AAS), the [Cd] u in samples from Mae Sot group was determined by 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 LOD of Cd quantitation, calculated as 3 times the standard deviation of blank measurements was 0.06 µg/L. None of the urine samples from this group was found to have a [Cd] u below the detection limit.

Estimated Glomerular Filtration Rate (eGFR)
The GFR is the product of nephron number and mean single nephron GFR, and in theory the GFR is indicative of nephron function [32][33][34]. In practice, the GFR is estimated from established chronic kidney disease-epidemiology collaboration (CKD-EPI) equations, and is reported as eGFR.
Male  [36]. Derivation of an equation for C cr normalization is in Appendix A. Demonstration that E Cd /C cr is unaffected by muscle mass is in Appendix B.

Benchmark Dose Computation and Benchmark Response (BMR) Setting
We used the web-based PROAST software version 70.1 (https://proastweb.rivm.nl (accessed on 9 June 2022) to compute the BMD figures for Cd exposure as E Cd /E cr or E Cd /C cr associated with three effect markers, eGFR, E NAG, and E β2M . A specific effect size, termed BMR, was set at 5% for all continuous endpoints, given that the BMDL values obtained could serve as the point of departure to derive health guidance values [25][26][27]. BMD values were computed from fitting datasets to multiple dose-response models such as inverse exponential, natural logarithmic, exponential, and Hill models. BMDL/BMDU values were computed from those showing a statistically significant dose-response relationship. Model averaging was used instead of a single model analysis to account for model uncertainty.
The BMR was set at a 10% increase in the prevalence of the following quantal endpoints; eGFR ≤ 60 mL/min/1.73 m 2 , E NAG /E cr ≥ 4 U/g creatinine, E β2MG /E cr ≥ 300 µg/g creatinine, E NAG /C cr × 100 ≥ 4 U/L filtrate, and E β2MG /C cr × 100 ≥ 300 µg/L filtrate. These endpoints were indicative of the presence of CKD, toxic injury to kidney tubular epithelial cells, and a reduction in tubular reabsorption of filtered protein, respectively. BMD values were calculated from fitting datasets to multiple dose-response models that included two-stage, logarithmic logistic, Weibull, logarithmic probability, gamma, exponential and Hill models.
The BMDL and BMDU corresponded to the lower bound and upper bound of the 95% confidence interval (CI) of BMD. The BMDL/BMDU were from model averaging using bootstrap with 200 repeats. A wider BMDL-BMDU difference, a higher statistical uncertainty in the dataset [28,29,37].

Statistical Analysis
Data were analyzed with IBM SPSS Statistics 21 (IBM Inc., New York, NY, USA). The distributions of eGFR and the excretion of Cd, β 2 M and NAG were examined for skewness, and those showing rightward skewing were subjected to logarithmic transformation before analysis. The departure of a given variable from a normal distribution was assessed with the one-sample Kolmogorov-Smirnov test. The Kruskal-Wallis test was used to compare mean differences across three eGFR groups. The Mann-Whitney U-test was used to compare mean differences between two groups. The Chi-square test was used to determine differences in percentage and prevalence data. For each analysis, p-values ≤ 0.05 for two-tailed tests were assumed to indicate statistical significance.

Characteristics of Males and Females of Low-and High-Exposure Locations
Among 734 participants, 100 males and 100 females were residents of a low-exposure area (Bangkok), and 189 males and 345 females were residents of a high-exposure area (Mae Sot) residents ( Table 1).
The overall mean age of 734 participants was 48.1 years. Of the total participants, 42.8% were smokers to include those who had quite less than 10 years. Another 31.7% of participants had hypertension and only 1.5% had diabetes. The overall % of CKD was 9%. None of subjects of the low-exposure group had CKD, while the % of CKD among males and females of the high-exposure group were 15.3% and 10.7%, respectively.
The overall % of subjects with a sign of tubular cell injury evident from E NAG /E cr ≥ 4 U/g creatinine was 76.2%, while the overall % of subjects showing a sign of tubular dysfunction was 39.8%.
Smoking was particularly high among males in the Mae Sot group (81.5%) than in males of the Bangkok group (47%). All females of the Bangkok group were non-smokers, while 32.8% of female residents of Mae Sot smoked cigarettes.
Mean E Cd for smokers in Mae Sot was 18.3-fold higher than that of smokers in Bangkok (5.87 vs. 0.32 µg/L, p < 0.001). The mean E Cd was 24.9-fold higher in non-smokers of the Mae Sot than non-smokers of Bangkok (5.22 vs. 0.21 µg/L, p < 0.001).
For the Bangkok group, mean [cr] p was lower in females than males (p < 0.001) as was mean E NAG /C cr (p = 0.023). For the Mae Sot group, mean [cr] p and mean [cr] u in females were lower than males (p < 0.001 for both parameters) as were mean [Cd] u (p = 0.020) and mean E Cd /C cr (p = 0.021). In contrast, mean E NAG /E cr in females was higher than that of males (p < 0.001).
The percentages of females and those with hypertension in each eGFR group did not differ, while % of diabetes (6.1%) and smoking (63.6%) were the highest in the lowest eGFR group. None of subjects from a low exposure area had low eGFR. Subjects of the low eGFR group was the oldest (mean age 63.2), who also had the lowest mean BMI [(p = 0.020). Mean [cr] p , mean [cr] u and mean [Cd] u all were the highest in the eGFR level 3, middle in eGFR level 2, and the lowest in eGFR level 3 as were the means for E cr -and C cr -normalized E Cd , E β2MG , E NAG .

BMDL and BMDU Values of Cadmium Excretion from E cr -Normalized Dataset
The BMDL and BMDU values of E Cd /E cr computed from three kidney toxicity endpoints; tubular cell injury, tubular dysfunction and eGFR reduction are provided separately for males and females (Table 3).  In continuous dose-response analyses, the respective BMDL/BMDU of E Cd /E cr associated with a tubular cell injury in men and women were 0.060/0.504 and 0.069/0.537 µg/g creatinine ( Figure S1), while the corresponding BMDL/BMDU of E Cd /E cr associated with a tubular dysfunction were 0.019/0.411 and 0.022/0.491 µg/g creatinine ( Figure S2). The BMDL (NOAEL equivalent) values of E Cd /E cr associated with these tubular toxicity indicators could be reliably determined, given that the BMDU/BMDL ratios ranged between 8.44 and 23. In contrast, the BMDU/BMDL ratios of E Cd /E cr associated with a 5% decrease in eGFR were extremely large ( Figure S3). Thus, the respective BMDL/BMDU values of E Cd /E cr associated with eGFR effect in men and women of 1.57 × 10 −6 /0.427 and 1.88 × 10 −5 /0.495 µg/g creatinine were not reliably determined.
In quantal dose-response analyses ( Figure S4), the respective BMDU/BMDL ratios of E Cd /E cr associated with a 10% increase in the prevalence of tubular cell injury in men and women were 1.48 × 10 −3 /0.56 and 3.17 × 10 −4 /0.706 µg/g creatinine. These BMDL values were not reliable due to the extremely large BMDL/BMDU ratios of E Cd /E cr in both men and women. The respective BMDL/BMDU values of E Cd /E cr associated with a 10% incidence of an increase in prevalence of tubular dysfunction were reliably determined as 0.469/0.973 and 0.733/1.29 µg/g creatinine. Likewise, reliable BMDL/BMDU values of E Cd /E cr associated with a 10% increase in the prevalence of CKD in men and women were 3.26/7.46 and 4.98/9.68 µg/g creatinine.

BMDL and BMDU Valuse of Cadmium Excretion from C cr -Normalized Dataset
The BMDL and BMDU values of E Cd /C cr computed from three kidney toxicity endpoints; tubular cell injury, tubular dysfunction and eGFR reduction are provided separately for males and females (Table 4).
In continuous dose-response analyses, the respective BMDL/BMDU of E Cd /C cr × 100 associated with a tubular cell injury in men and women were 0.067/0.394, and 0.067/0.399 µg/L filtrate ( Figure S5), while the corresponding BMDL/BMDU of E Cd /C cr × 100 associated with a 5% reduction in reabsorption of β 2 M were 0.41 × 10 −4 /0.038, and 0.16 × 10 −3 /0.040 µg/L ( Figure S6). The BMDL (NOAEL equivalent) values of E Cd /C cr × 100 associated with a defective tubular reabsorption in men and women could not be reliably determined, due extremely large BMDU/BMDL ratios, indicative of a high degree of statistical uncertainty. In contrast, the BMDU/BMDL values of E Cd /C cr × 100 associated with a 5% eGFR decline were determined with a sufficiently high degree of statistical certainty as 0.080/0.714 and 0.098/0.812 µg/L filtrate in men and women, respectively ( Figure S7). In quantal dose-response analyses ( Figure S8), the respective BMDU/BMDL ratios of E Cd /C cr associated with a 10% increase in the prevalence of tubular cell injury in men and women 2.54/4.58, and 3.66/6.14 µg/L filtrate, respectively. The respective BMDL/BMDU values of E Cd /C cr × 100 associated with a 10% incidence of an increase in prevalence of defective tubular dysfunction were 0.456/0.807 and 0.5/0.805 µg/ L filtrate, respectively. Likewise, reliable BMDL/BMDU values of E Cd /C cr × 100 associated with a 10% increase in the prevalence of CKD in men and women were 5.06/8.66 µg/ L filtrate in men and 5.15/7.62 µg/L filtrate in women.

Discussion
To the best of our knowledge, we herein provide, for the first time, BMDL values of Cd excretion levels that are derived simultaneously from three toxicity endpoints; injury to kidney tubular epithelial cells, a defective tubular reabsorption of β 2 MG, and eGFR decline. As age distribution histogram indicates ( Figure S9), these BMDL values of permissible Cd accumulation levels can be generalized to an entire adult population. A Cd accumulation level producing E Cd /C cr of 0.67 ng/L filtrate in men and women could be considered as Cd accumulation levels below which renal effects are likely to be negligible ( Figure S5). A reduction in eGFR may follow when Cd accumulation levels rise to 0.80 ng/L filtrate in men and 0.98 ng/L filtrate in women ( Figure S7).
The reported BMD values of E Cd /E cr associated with adverse effects on kidneys vary among studies. The variability was most likely due to differences in age of subjects included in an analysis as well as differences in the dose-response models applied to the datasets [25][26][27]. It is noteworthy that most BMD values of E Cd associated with adverse kidney effects were derived from those age 50 years or older. Consequently, most BMD figures previously published were not generalizable to the entire adult population. For a quantal endpoint, different cut-off values constituted an additional source of variable BMD values. There is a need to standardize the BMD approach to advance Cd toxicological risk assessment. Age should form an important consideration in deriving BMDL.
In the present study, we used E NAG /E cr of 4 U/g creatinine as the cut-off value to reflect injury to kidney tubular epithelial cells ( Figure S1). E NAG is considered to be proportional to the number of surviving nephrons because NAG in urine originated from injured or dying tubular cells [38][39][40]. In a United Kingdom (U.K.) study, a dose-response relationship was observed between E Cd and E NAG ; E Cd /E cr of 0.5 µg/g creatinine was associated with 2.6-and 3.6-fold increases in the prevalence of E NAG /E cr > 2 U/g creatinine, as compared with a E Cd /E cr of 0.3 and < 0.5 µg/g creatinine, respectively [41]. The BMDL/BMDU figures derived by us from E NAG endpoint were in line with the U.K. study. Apparently, these NOAEL equivalents of E Cd /E cr associated with tubular epithelial cell injury were lower than E Cd /E cr of 5.24 µg/g creatinine, suggested to be a nephrotoxicity threshold level of Cd [18].
The BMDL values of E Cd /E cr derived from kidney tubular cell injury endpoint were 0.060 and 0.069 µg/g creatinine in men and women, respectively ( Figure S1). The respective BMDL values of E Cd /E cr associated with a 5% decrease of tubular reabsorption of β 2 M in men and women are 0.019 and 0.022 µg/g creatinine ( Figure S2). In contrast, the BMDL values of E Cd /E cr associated with a 5% eGFR decline could not be reliably determined (U/L ratios of 10 4 -10 5 ). These findings were consistent with data from a conventional dose-response analysis, where an inverse relationship between E Cd and eGFR became statistically significant only when E Cd was normalized to C cr [42].
Because of high variance in datasets introduced by E cr -normalization, the effect of Cd on eGFR was not realized. In a systematic review and meta-analysis of pooled data from 28 studies, the risk of proteinuria was increased by 1.35-fold when comparing the highest vs. lowest category of Cd dose metrics, but an increase in the risk of low eGFR was statistically insignificant (p = 0.10) [43]. An erroneous conclusion that chronic Cd exposure was not associated with a progressive eGFR reduction was also made in another systematic review [44].
In a meta-analysis of pooled data from 30 publications, the BMD of E Cd /E cr for tubular injury (E NAG ) was 1.67 µg/g creatinine [45]. In another study using data from 469 men and 465 women who were resident of Zhejiang Province, China, the E Cd /E cr BMD/BMDL figures based on E β2M and E NAG were 1.24/0.62 and 0.85/0.49 µg/g creatinine in men, respectively [46]. The corresponding figures for women were 1.35/0.64 and 1.36/0.65 µg/g creatinine, respectively. Of note, in men only, E Cd /E cr BMD/BMDL figures for E β2MG were higher than those for E NAG . This was seen in a Swedish study [47] and our datasets.
In a study of data from 410 men and 418 women, aged 40-59 years, who were residents of non-pollution areas of Japan, the E Cd /E cr BMD figures for E β2M endpoint were 0.6-1.2 and 0.6-2.3 µg/g creatinine in men and women, respectively [48].
According to a quantal endpoint of E β2M /E cr ≥ 300 µg/g creatinine, the BMDL value E Cd /E cr values were 0.469 and 0.733 µg/g creatinine in men and women, respectively ( Figure S4). These data imply that the prevalence of Cd-induced tubular dysfunction was likely to be smaller than 10% at E Cd /E cr 0.469 µg/g creatinine in men, and 0.733 µg/g creatinine in women. Of concern, these E Cd /E cr levels have been associated with increased risk of CKD in the general populations in the U.S. [49] Spain [50], Taiwan [51] and China [52,53]. Arguably, E β2M is not a sensitive toxicity endpoint and consequently the dietary exposure guideline of tolerable intake at 0.83 µg/kg body weight/day derived from E β2M /E cr ≥ 300 µg/g creatinine endpoint does not afford sufficient health protection.
The BMD values for E NAG and eGFR reduction in a study of 790 Swedish women, aged 53-64 years, were 0.5-0.8 and 0.7-1.2 µg/g creatinine, respectively [47]. A Cd-induced kidney tubular cell injury and eGFR reduction could be more suitable endpoints for health risk calculation than E β2M . In clinical trials, the successful treatment of kidney disease is judged by the attenuation of a decline in eGFR [33,54]. CKD is a progressive syndrome with high morbidity and mortality, and affects 8% to 16% of the world's population. The continuing rise in the incidence of CKD globally is a cause of concern [54].
It has long been viewed that excreted Cd included Cd molecules that pass through the glomerular filtration membrane into the filtrate but are not reabsorbed. However, it is noteworthy that the excreted Cd originates from injured or dying tubular cells [40]. In an analogy to E NAG , Cd excretion is a manifestation of the cytotoxicity of its accumulation in kidneys' tubular cells. In a histopathological examination of kidney biopsies from healthy kidney transplant donors [55], the degree of tubular atrophy was positively associated with the level of Cd accumulation. Tubular atrophy was observed at relatively low Cd levels [55].
The reductions in eGFR due to Cd nephropathy have often been attributed to glomerular injury. However, current evidence suggests that sufficient tubular injury disables glomerular filtration and leads to nephron atrophy and GFR loss [40,56]. Cd inflicts tubular cell injury at low intracellular concentrations, and the toxicity intensifies as Cd concentration rises [40]. The Cd-induced tubular injury disables glomerular filtration, leading to nephron atrophy, glomerulosclerosis, and interstitial inflammation and fibrosis. A reduction in tubular reabsorption of filtered proteins, retinal binding protein (RBP) and β 2 MG follows tubular atrophy and nephron loss. An increase of β 2 M excretion to levels above 300 µg/g creatinine is speculatively due to effects of Cd on both tubular reabsorption and nephron number. Figure 1 depicts the pathogenesis of Cd-induced nephropathy that originates from tubular injury deduced from the BMDL values of E Cd /C cr provided in Table 4. In theory, an acceptable level of environmental exposure to Cd should be derived from the most sensitive toxicity endpoint, which is the one with the lowest BMDL value [27]. In the present study, BMDL for a 5% increase of E β2M /C cr could not be reliably determined (Table 4), thereby arguing that the utility of E β2M as a toxicity criterion is not appropriate. The BMDL value of E Cd for E NAG endpoint appeared to be the lowest (Table 4). Thus, a tolerable intake level of Cd derived from E NAG /C cr endpoint will be sufficiently low to protect against Cd-induced toxic tubular cell injury. This E NAG -based tolerable intake level of Cd will be lower than a tolerable intake level of 0.83 µg/kg body weight per day that was derived from E β2M endpoint.
The protein β 2 M, with a molecular weight of 11,800 Da, is synthesized and shed by all nucleated cells in the body [57]. Owing to its small mass, β 2 M is filtered freely by the glomeruli and is reabsorbed almost completely by the kidney's tubular epithelial cells [58][59][60]. Increased β 2 M excretion has been used as an indicator of impaired tubular re-absorptive function. It is noteworthy that β 2 M production rises in response to many inflammatory and neoplastic conditions [61] and this compromises its utility as a marker of tubulopathy. If reabsorption rates of β 2 M per nephron remain constant as its production rates change, excretion will vary directly with its production [Appendix C]. If the production and reabsorption per nephron remain constant as nephrons are lost, the excretion of β 2 M will rise [Appendix D].
The reductions in eGFR due to Cd nephropathy have often been attributed to glomerular injury. However, current evidence suggests that sufficient tubular injury disables glomerular filtration and leads to nephron atrophy and GFR loss [40,56]. Cd inflicts tubular cell injury at low intracellular concentrations, and the toxicity intensifies as Cd concentration rises [40]. The Cd-induced tubular injury disables glomerular filtration, leading to nephron atrophy, glomerulosclerosis, and interstitial inflammation and fibrosis. A reduction in tubular reabsorption of filtered proteins, retinal binding protein (RBP) and β2MG follows tubular atrophy and nephron loss. An increase of β2M excretion to levels above 300 μg/g creatinine is speculatively due to effects of Cd on both tubular reabsorption and nephron number. Figure 1 depicts the pathogenesis of Cd-induced nephropathy that originates from tubular injury deduced from the BMDL values of ECd/Ccr provided in Table 4. Figure 1. Sequential outcomes of tubular cell toxic injury of cadmium accumulation in kidneys. Cd inflicts tubular cell injury at low intracellular concentrations, and the toxicity intensifies as Cd concentration rises [40]. Tubular injury disables glomerular filtration, causing nephron atrophy, glomerulosclerosis, and interstitial inflammation and fibrosis [40,55,56]. A reduction in tubular reabsorption of filtered proteins, RBP and β2MG follows tubular atrophy and a substantial nephron loss. Abbreviation: KIM1, kidney injury molecule 1; NAG, N-acetyl-β-D-glucosaminidase, NGAL, neutrophil gelatinase associated lipocalin; RBP, retinal binding protein; β2M, β2-microglobulin. In theory, an acceptable level of environmental exposure to Cd should be derived from the most sensitive toxicity endpoint, which is the one with the lowest BMDL value [27]. In the present study, BMDL for a 5% increase of Eβ2M/Ccr could not be reliably determined (Table 4), thereby arguing that the utility of Eβ2M as a toxicity criterion is not appropriate. The BMDL value of ECd for ENAG endpoint appeared to be the lowest (Table 4). Thus, a tolerable intake level of Cd derived from ENAG/Ccr endpoint will be sufficiently low to protect against Cd-induced toxic tubular cell injury. This ENAG-based tolerable intake level of Cd will be lower than a tolerable intake level of 0.83 μg/kg body weight per day that was derived from Eβ2M endpoint.
The protein β2M, with a molecular weight of 11,800 Da, is synthesized and shed by all nucleated cells in the body [57]. Owing to its small mass, β2M is filtered freely by the Cd inflicts tubular cell injury at low intracellular concentrations, and the toxicity intensifies as Cd concentration rises [40]. Tubular injury disables glomerular filtration, causing nephron atrophy, glomerulosclerosis, and interstitial inflammation and fibrosis [40,55,56]. A reduction in tubular reabsorption of filtered proteins, RBP and β 2 MG follows tubular atrophy and a substantial nephron loss. Abbreviation: KIM1, kidney injury molecule 1; NAG, N-acetyl-β-D-glucosaminidase, NGAL, neutrophil gelatinase associated lipocalin; RBP, retinal binding protein; β 2 M, β 2 -microglobulin. Consistent with the above notions are data in Table 2 where mean E β2M /C cr in those with low eGFR was 90-fold higher than those with eGFR ≥ 90 mL/min/1.73 m 2 , while there was only a 3-fold increase in E β2M /C cr in those with moderate eGFR (eGFR 61-89 mL/min/1.73 m 2 ). E β2M is a function of nephron numbers and tubular reabsorption activity, a substantial increase in E β2M can be expected when nephrons are lost [24]. Accordingly, the sue of E β2M as a "critical" effect of Cd accumulation in kidney tubular cells is inappropriate.

Conclusions
The BMDL/BMDU values of E Cd /C cr associated with a 5% increase of kidney tubular cell injury are 0.67/3.94 and 0.67/3.99 ng/L filtrate in men and women, respectively. The respective BMDL/BMDU values of E Cd /C cr associated with a 5% eGFR decline in men and women are 0.80/7.14 and 0.98/8.12 ng/L filtrate. A Cd accumulation level resulting in Cd excretion at 0.67 ng/L filtrate in men and women could be viewed as NOAEL equivalent and is a maximally permissible Cd accumulation level. A reduction in eGFR and CKD may follow when Cd excretion levels exceed this maximal permissible accumulation level of 0.67 µg/L of filtrate. Health guidance values for Cd derived from these NOAEL equivalents of Cd excretion levels are more likely to be low enough to carry a negligible adverse effect on kidneys.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph192315697/s1, Figure S1. Continuous dose-response analysis of E Cd /E cr and E NAG /E cr . Datasets were fitted to 4 dose-response models; an exponential model (a), Hill model (b), an inverse exponential model (c), and natural logarithmic model (d).
Model averaging BMDL/BMDU values of E Cd /E cr associated with a 5% increase in E NAG /E cr (e); Figure S2. Continuous dose-response analysis of E Cd /E cr and E β2MG /E cr . Datasets were fitted to an exponential model (a), and Hill model (b). Model averaging BMDL/BMDU values of E Cd /E cr associated with a 5% increase in E β2MG /E cr (c); Figure S3. Continuous dose-response analysis of E Cd /E cr and eGFR. Datasets were fitted to 4 dose-response models including, an exponential model (a), Hill model (b), an inverse exponential model (c), and natural logarithmic model (d).
Model averaging BMDL/BMDU values of E Cd /E cr associated with a 5% decrease in eGFR (e); Figure S4. Quantal dose-response analysis of E Cd /E cr . Bootstrap curves of model averaging of E Cd /E cr values associated 10% increases in the prevalence of tubular cell injury (a), tubular dysfunction (c), and CKD (e). BMDL and BMDU values of E Cd /E cr according to tubular injury (b), tubular dysfunction (d), and CKD (f); Figure S5. Continuous dose-response analysis of E Cd /C cr and E NAG /C cr . Datasets were fitted to 4 dose-response models; an exponential model (a), Hill model (b), an inverse exponential model (c), and natural logarithmic model (d). Model averaging BMDL/BMDU values of E Cd /C cr associated with a 5% increase in E NAG /C cr (e); Figure S6. Continuous dose-response analysis of E Cd /C cr vs. E β2MG /C cr . Datasets were fitted to an exponential model (a), and Hill model (b). Model averaging BMDL/BMDU values of E Cd /C cr associated with a 5% increase in E β2MG /C cr (c); Figure S7. Continuous dose-response analysis of E Cd /C cr vs. eGFR. Datasets were fitted to 4 dose-response models; an exponential model (a), Hill model (b), an inverse exponential model (c), and natural logarithmic model (d). Model averaging BMDL/BMDU values of E Cd /C cr associated with a 5% decrease in eGFR (e); Figure S8. Quantal dose-response analysis of E Cd /C cr . Bootstrap curves of model averaging of E Cd /C cr values associated 10% increases in prevalence of tubular cell injury (a), tubular dysfunction (c), and CKD (e). BMDL and BMDU values of E Cd /C cr corresponded to 10% increases in prevalence of tubular injury (b), tubular dysfunction (d), and CKD (f); Figure S9. Histogram of age distribution of study subjects. There were 289 men and 445 women drawn from Bangkok (a low-exposure area) and from a Cd contaminated area of Mae Sot District (a high-exposure area) of Thailand [30]. The overall percentages of smokers, hypertension, and diabetes were 42.8%, 31.7% and 1.5%, respectively. The overall % of subjects with evidence for CKD, tubular cell injury and tubular dysfunction were 9%, 76.2%, and 39.8%, respectively. Institutional Review Board Statement: This study analyzed archived data taken from a published report [30]. Ethical review and approval were not applicable.

Informed Consent Statement:
All participants took part in the study after giving informed consent. Data Availability Statement: All data are contained within this article.
Acknowledgments: This work was supported with resources from the Kidney Disease Research Collaborative, and the Department of Nephrology, Princess Alexandra Hospital. It was also supported with the resources of the Department of Toxicology "Akademik Danilo Soldatović", University of Belgrade-Faculty of Pharmacy, Serbia.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A
An equation for Normalizing Excretion of x (E x ) to Creatinine Clearance (C cr ) [40] Let V u = urine flow rate, units of volume/time; E x = urinary excretion rate of substance x, units of mass/time; [x] u = urinary concentration of substance x, units of mass/volume; E cr = urinary excretion rate of creatinine, units of mass/time;

Appendix C
Demonstration that E β2M May Rise because of Increased Endogenous Production [40] Let I β2M = influx of β 2 M from endogenous sources into plasma; [β 2 M] p = plasma concentration of β 2 M, mg/L; F β2M = rate of glomerular filtration of β 2 M, mg/d; E β2M = urinary excretion rate of β 2 M, mg/d; TD β2M = rate of tubular degradation of β 2 M, mg/d; GFR = glomerular filtration rate, L/d. Assume that F β2M = GFR[β 2 M] p = E β2M + TD β2M . Assume an equilibrium between Iβ 2 M and Fβ 2 M, and assume stable GFR. If I β2M rises, so do [β 2 M] p and F β2M . If TD β2M remains stable, E β2M must rise if F β2M rises. Thus E β2M may rise even though TD β2M has not fallen.

Appendix D
Demonstration that E β2M Rises if GFR Falls [40] From Appendix C, F β2M = GFR[β 2 M] p = E β2M + TD β2M . Dividing the equation on the right by GFR, [β 2 M] p = E β2M /GFR + TD β2M /GFR. Assume that I β2M and thus F β2M remain stable. Since F β2M = GFR[β 2 M] p , [β 2 M] p must rise reciprocally if GFR falls. Assume that TD β2M /GFR remains stable as GFR falls. If GFR has fallen, TD β2M must also fall. Since F β2M is constant and F β2M = E β2M + TD β2M , E β2M must rise.