Cadmium exposure is associated with chronic kidney disease in a superfund site lead smelter community in Dallas, Texas

Abstract

Background: The study was conducted in a Dallas lead smelter community following an Environmental Protection Agency (EPA) Superfund Cleanup project. Lead smelters operated in the Dallas community since the mid-1930s.Aim: To test the hypothesis that cadmium (Cd) exposure is associated with chronic kidney disease (CKD) ≥ stage 3.Subjects and methods: Subjects were African American residents aged ≥19 to ≤ 89 years (n=835). CKD ≥ stage 3 was predicted by blood Cd concentration with covariates.Results: In logistic regression analysis, CKD ≥ stage 3 was predicted by age ≥ 50 years (OR = 4.41, p < 0.0001), Cd level (OR = 1.89, p < .05), hypertension (OR = 3.15, p < 0.03), decades living in the community (OR = 1.34, p < 0.003) and T2DM (OR = 2.51, p < 0.01). Meta-analysis of 11 studies of Cd and CKD ≥ stage 3 yielded an ORRANDOM of 1.40 (p < 0.0001). Chronic environmental Cd exposure is associated with CKD ≥ stage 3 in a Dallas lead smelter community controlling covariates.Conclusion: Public health implications include screening for heavy metals including Cd, cleanup efforts to remove Cd from the environment and treating CKD with newer renal-sparing medications (e.g., SGLT-2 inhibitors, GLP-1s).

Keywords:

• Cadmium
• chronic kidney disease
• African American
• EPA superfund

1. Introduction

Heavy metals are well-known environmental toxins, some of which are associated with chronic kidney disease (CKD). With the rapid increase in industrialisation over the past 100 years, trace metals have become an increasingly important source of pollution (Khan et al. 2021). This has resulted in a dramatic and increasing human-induced mobilisation of trace metals into the biosphere. Near the end of the last century, it was estimated that each year 1,160,000 tons of lead, 120,000 tons of arsenic, 30,000 tons of cadmium, and 11,000 tons of mercury were added to the biosphere (Khan et al. 2021). Worldwide toxicity of these “new” trace metals now exceeds the toxicity of all radioactive and organic wastes generated yearly.

Cd environmental contamination results from rechargeable nickel-cadmium batteries, smelting, mining, household waste incineration, fossil fuel combustion, and phosphate-containing fertilisers (Elinder and Järup 1996). Diet is the most important source of cadmium in non-smokers because of soil or water contamination. Greater than 80% of dietary Cd comes from contaminated cereals, vegetables and potatoes (Olsson et al. 2002). Cd also enters the body via the lung in cigarette smokers and with occupational exposure. Cd concentrates in tobacco plants, where the soil to plant transfer rate is very high. Smokers have kidney [Cd] 2–3.5-fold higher than non-smokers (Nilsson et al. 1995; Satarug et al. 2002), The Agency for Toxic Substances and Disease Registry has estimated that more than 500,000 workers in the United States are occupationally exposed to Cd primarily from dust and airborne particles from smelter activity (Jaishankar et al. 2014).

Cd absorbed in the GI tract is transported via the portal system to liver, where it binds metallothionein (MT). The Cd mode of toxicity is through its action as a free ion that is tightly bound by MT, which acts as a detoxifying mechanism (Satarug 2018). Cd-MT complexes also form in lung. Cd-MT is released into the systemic circulation where it is taken up by proximal tubular cells. Given the long biological half-life of Cd in kidney, 7–45 years, it remains there for decades (Järup et al. 1983; Fransson 2014). The Cd-MT complex enters endosomes and lysosomes in proximal tubule where proteases degrade MT releasing free Cd into the cytoplasm. Toxicity at a cellular level results from generation of reactive oxygen species, interference with normal DNA repair mechanisms and induction of apoptosis (Järup and Åkesson 2009).

The most common and earliest presentation of Cd-associated kidney disease is proximal tubular damage most frequently manifested by an increase in urinary excretion of low molecular weight (LMW) proteins such as β2-microglobulin (Järup and Åkesson 2009). LMW proteinuria is the subject of the majority of studies examining renal Cd effects. More severe tubular injury can result in phosphate wasting, aminoaciduria, renal glycosuria, decreased concentrating ability, and hypercalciuria. Hypercalciuria may be responsible for the increased incidence of nephrolithiasis reported with occupational Cd exposure (Järup and Elinder 1993). Interstitial inflammation and fibrosis can lead to glomerular damage, reduction in glomerular filtration rate, end-stage kidney disease (ESKD) and increased mortality (Nishijo et al. 2017). However, the association of Cd with significant glomerular disease (CKD ≥ stage 3) is less well described with some studies detecting increased risk (Ferraro et al. 2010; Navas-Acien et al. 2009; Chung et al. 2014; Lin et al. 2014; Wu et al. 2019; Lee et al. 2020; Yuan et al. 2020). Two studies found no association of Cd with CKD (Sommar et al. 2013; Tsai et al. 2021), and still others found increased risk in subpopulations such as women (Myong et al. 2012) and those with diabetes mellitus or hypertension (Kim et al. 2015). Knowledge gaps in the association of Cd exposure and CKD include: timing and magnitude of exposure, longitudinal data, ascertainment of exposure, CKD definition, and appropriate statistical methods. CKD definitions based upon laboratory eGFR values were not always used in prior studies, and attention must be given to appropriate treatment of chemical concentrations (eg transform to log₁₀ values) (Moody et al. 2018, p. xxx). The present investigation provides known individual level exposures at a known time, and years of residence in the community (ie length of exposure). The final gap in Cd studies is lack of longitudinal studies. Unfortunately, a cross-sectional study that includes only one visit cannot address the need for serial data.

We previously reported an association of chronic environmental and occupational lead exposure with worsening kidney function in a group of African American residents in Dallas, Texas, exposed to lead smelters (Reilly et al. 2018). The smelters, which reprocessed Pb batteries, contaminated the environment with lead and Cd. Subsequently, we also reported an association between type II diabetes mellitus (T2DM) and Cd exposure in this same population (Little et al. 2020). In the present manuscript, we extend our previous studies and examine the relationship between Cd exposure and CKD ≥ stage 3. We also performed a meta-analysis of published literature examining the association between Cd and CKD ≥ stage 3.

2. Subjects and methods

The present study was evaluated by the Institutional Review Board at University of Texas Southwestern Medical Centre, Dallas, Texas and approved (IRB#0902-495, 4 September 2002) The study adheres to the Declaration of Helsinki.

2.1. Study population

Three lead smelters opened in 1934 in West Dallas (RSR). The first one was established in 1934. Two other lead smelters began operation near each other in East Dallas in the 1940s, in Cadillac Heights. Environmental lead was a known public health hazard in the mid-1960s. The City of Dallas Health Department began operating lead clinics for walk-in testing and conducted health surveys. Several lead remediation projects were done between 1970 and 1992, including soil removal. Blood lead levels in adults and children decreased only after the smelters closed. Smelting lead from slag and batteries released dust particle clouds over these communities. The West Dallas smelter was several blocks south of one of the city’s largest public housing developments. The last smelter closed in 1990. The West Dallas lead smelter area was designated an US Environmental Protection Agency (EPA) Superfund Site. The Superfund cleanup included topsoil removal from the area and renovation or demolition of old or dilapidated housing. New public housing was built in the old smelter communities after environmental remediation.

2.2. Study design

The study population is a convenience sample made up of African Americans living in postal codes where the smelters operated (1946–1984). In the total sample there were seven whites, 46 Hispanics, and three “Other race” individuals. They were excluded because there was such a small number, and excluding them achieved high homogeneity of ethnicity of individuals included. Participants were recruited through town hall meetings and public service media announcements in the media (radio, television and newspapers). People living in the following zip codes were included in the study sample: 75203, 75208, 75211, 75212, 75215, 75216, and 75247. The vast majority were from the West Dallas smelter community. Physical examinations and clinical laboratory tests were conducted. Inclusion criteria were having valid values listed for age, sex, creatinine and race, for eGFR calculation; BMI, blood pressure, HbA1c, GGT concentration, tobacco smoking status, heavy metal levels (Cd, Pb, Ar, Hg) and lived in a smelter area zip code. Eight hundred and seventy-five individuals met these criteria. Of 875 individuals who met inclusion criteria, 55 were male smelter-working residents and 820 were non-smelter working residents. Fourteen female smelter-working residents were excluded because of missing values. Hypertension (blood pressure ≥140/90 mmHg) was dummy coded (0,1) for regression analysis. Smoking was coded current smoker (one-half pack per day of cigarettes or more =1) or non-smoker (value =0). T2DM was defined with an HbA1c value of 6.5%, as recommended by the American Diabetes Association (American Diabetes Association 2020). T2DM (HbA1c ≥6.5%) was dummy coded (0, 1) for regression analysis. BMI was analysed as a continuous variable because all individuals in the study were obese (BMI >30).

2.3. Estimated glomerular filtration rate (eGFR)

Estimated glomerular filtration rate (eGFR) was calculated using the MDRD (Modification of Diet in Renal Disease) equation for African Americans: male eGFR = 186 × serum creatinine^−1.154×age^−0.203×1.21 and female eGFR = 186 × serum creatinine^−1.154×age^−0.203×1.21 × 0.742 (Levey et al. 1999). The MDRD equation was used to allow comparison results to previous studies that also used this equation. Due to rightward skewing, blood Cd level was log₁₀ transformed (>1.0) and is the usual transform for chemical concentrations.

2.4. Analytical techniques

Logistic regression was used to analyse demographics and heavy metal exposure to prediction of CKD ≥ stage 3 status. A systematic literature review was done. Meta-analysis of studies that met inclusion criteria (reported primary data, sample sizes, Cd blood or urine levels, method of Cd level determination, and a generally recognised endpoint for CKD ≥ stage 3) was also conducted. For inclusion, a study included a recognised endpoint, eGFR ≤60 ml/min/1.73m² (CKD ≥ Stage 3), which was calculated using either the MDRD or CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equations, dialysis or renal transplantation.

The dependence of CKD ≥ stage 3 (0,1) on independent variables was analysed with logistic regression with the null hypothesis that CKD ≥ stage 3 is not dependent on exposure to Cd or other metals. The regression model was: CKD ≥ stage 3 (0,1) = age >50 years (0,1) + sex: male (0,1) + smelter worker (0,1) + log₁₀ (blood Cd level) + log₁₀ (blood arsenic level) + log₁₀ (blood mercury level) + log₁₀ (blood lead level) + duration of residence (decades) + T2DM (0,1) + hypertension (0,1) + T2DM (0,1) + smoking tobacco (0,1) + abnormal gamma-glutamyl-transpeptidase (GGT) (0,1), where 0 = No, 1 = Yes. Software SAS (v.9.1, SAS Institute, Cary, NC, USA) and IBM SPSS (v.27, IBM, Chicago, IL, USA) were used for data analysis. Arsenic was ultimately dropped from the analysis because it caused multicollinearity.

2.5. Meta-analysis

Meta-analysis using Comprehensive Meta-Analysis, Version 3 (Biostat, Englewood, NJ, USA) was done. Of the studies that reported CKD and cadmium exposure (n = 125), we included only those published studies in which individual exposures were documented and analysed (n = 11, see Table 3A). Studies that used an ecological approach (see Table 3B) were excluded because there is no documentation of an individual’s exposure. Such studies are confounded by the ecological fallacy (ie associating group characteristics to individuals without individual level data) (Schwartz 1994). Studies were also excluded from the meta-analysis for non-standard reporting of renal function (Haswell-Elkins et al. 2008), different CKD measures (Mason et al. 1988; Boonprasert et al. 2018), estimated Cd intake based on dietary intake without quantifying Cd in the food (Thomas et al. 2014; Shi et al. 2018; Wang et al. 2021), and risk based on estimated exposure but no actual measurement of Cd concentration (Hellström et al. 2001).

Table 3. Selected studies examining the relationship between Cd and CKD ≥ stage 3/ESKD (exposure based on blood or urine [Cd] [A] and dietary surveys or proximity to battery plants (ie ecological fallacy) [B]).

A
Author	Population	Study	CKD measure	Result
Ferraro 2010 (15)	NHANES 1999–2006	Cross-sectional N = 5426	CKD-EPI (GFR only)	Blood [Cd] > 1 mcg/L associated with eGFR <60 ml/min/1.73m^2, OR: 1.48 (95% CI, 1.02,2.17)
Navas-Acien 2009 (18)	NHANES 2007–2012	Cross-sectional N = 14,778	MDRD (GFR only)	Highest vs lowest quartile blood Cd associated with eGFR <60 ml/min/1.73m^2, OR: 1.32 (95% CI, 1.04,1.68), interaction with smoking, lead and HTN, lead and Cd co-exposure a strong determinant of CKD
Lin et al. 2014 (52)	NHANES 2011–2012 age >20	Cross-sectional N = 1545	CKD-EPI (GFR only), same result MDRD	Blood [Cd] > 0.53 mcg/L vs blood [Cd] < 0.18 mcg/L associated with eGFR <60 ml/min/1.73m^2, OR- 2.21 (95% CI, 1.09,4.5), low serum [Zn] increases risk
Lee 2020 (40)	NHANES 1999–2016	EWAS N = 46,748	CKD-EPI (GFR only)	Divided into discovery and validation sets, discovery set: OR 1.30 (95% CI, 1.19–1.42), validation set: OR 1.20 (95% CI, 1.1,1.3). Meta-analysis discovery set: OR: 1.3 (95% CI, 1.19,1.42), validation set: OR: 1.20 (95% CI, (1.1,1.3)
Kim et al. 2015 (24)	KNHANES 2011	Cross-sectional N = 1797	MDRD	Blood Cd. No association total population, association in those with HTN or DM, HTN: OR 1.52 (95% CI, 1.05,2.19), DM: OR 1.92 (95% CI, 1.14–3.25) with eGFR <60 ml/min/1.73m^2, Cd may contribute to CKD risk in those with HTN or DM
Chung 2014 (14)	KNHANES 2008	Cross-sectional N = 8641	CKD-EPI (GFR only)	Blood [Cd] quartiles Q4 vs Q1–3, OR 1.467 (95% CI, 1.077,1.999) associated with eGFR <60/ml/min/1.73m^2.
Kim and Lee 2012 (45)	KNHANES 2008–2010	Cross-sectional N = 5924	MDRD/not <60 ml/min/1.73m^2	Log transformed [Cd] quartiles OR: 0.806 (95% CI, 0.602,1.08. Non-standard definition for reduced renal function, 25^th percentile, cut-off <80.63 ml/min/1.73m^2
Myong et al. 2012 (23)	KNHANES 2005–2008	Cross-sectional N = 2992	MDRD	Whole blood Cd quartiles, compared Q4 vs Q1, association with eGFR <60 ml/min/1.73m^2 only in women OR: 2.48 (95% CI, 1.74,3.54), men OR: 0.74 (0.5,1.1)
Hwangbo 2011 (44)	KNHANES 2005, age >20	Cross-sectional N = 1909	CKD-EPI/not <60 ml/min/1.73m^2	Non-standard definition for reduced renal function, 25^th percentile using sex-specific cut-offs, men- <74.7 ml/min/1.73m^2, women <65.4 ml/min/1.73m^2
Mason 1988 (29)	UK occupational exposure	75 exposed, 75 matched controls	Cockcroft-Gault corrected for BSA	13/75 exposed with C-G < 50 ml/min/1.73m^2, 3/74 age and sex matched controls with C-G < 50 ml/min/1.73m^2, Chi-square p < .009
Wu 2019 (19)	Taiwan	N = 220 CKD, N = 438 controls	MDRD (GFR only)	Age and sex matched controls. RBC [Cd] tertiles, <0.80 mcg/L: referent, 0.8–1.3 mcg/L: OR 2.4 (95% CI: 1.14–5.07), >1.3 mcg/L: OR 8.77 (95% CI, 5.13–14.98)
Tsai 2021 (22)	Taiwan	Cross sectional N = 2447	CKD-EPI (GFR only)	Participants invited to take health survey. Urine [Cd] (log per 1 mcg/L) multivariate analysis, OR: 0.758 (95% CI, 0.404,1.423)
Kaewnate 2012 (46)	Thailand	Cross sectional	Cockcroft-Gault	Urinary Cd, multiple logistic regression OR: 3.73 (95% CI, 2.5,5.57) contaminated (n = 563) vs non-contaminated (n = 522) villages
Wang 2016 (47)	China Jiangshan City	Random sample N = 934	None	Kidney disease excluded. Looked at urine Cd and LMW proteinuria. No measure of eGFR
Anupama 2019 (53)	South India	Case-control	MDRD	69 matched CKD cases and controls, ESKD excluded. Cd in blood detected in one case and 5 controls. OR: 0.16 (95% CI, 0.01,1.12) not significant
Yuan 2020 (20)	Taiwan, residents near petrochemical plant	Exposure cohort N = 2069	CKD-EPI Taiwan	Urine Cd. High exposure within 10 km of plant, low exposure >10 km from plant. % change in eGFR for one-fold increase in exposure. OR: 0.96 high vs low exposure (95% CI 0.87,1.06) listed as significant, but apparently not.
Sommar 2013 (21)	Sweden	Matched case-control	ESKD, GFR <10–15 ml/min, dialysis, transplant	118 ESKD cases and 347 matched controls, matched for sex, age and area, time of blood sample. Erythrocyte Cd mcg/L, OR: 1.17 (95% CI, 0.99,1.38)
B
Author	Population	Study	CKD measure	Result
Shi 2018 (30)	China Health Nutrition Survey N = 8429	Nutritional survey: cross sectional	MDRD (GFR only) No difference MDRD	Quartiles of estimated Cd intake based on dietary survey. Highest vs lowest quartile OR: 4.05 (95% CI, 2.91–5.43). OR higher in those with DM (8.10) and HTN (7.85).
Hellström 2001 (33)	Kalmar County Sweden, age 20–79, resident 1978–1995	Retrospective cohort, occupational and residential exposure	ESKD or transplant	Risk stratification based on an exposure index based on proximity to Ni-Cd battery plants, high: worker > 1 year n = 655, moderate: within 2 km of plant n = 8825, low: 2–10 km away n = 7200, unexposed: >10 km, n = 152,477. Age and sex standardised Mantel–Haenszel rate ratio 1.8 (95% CI, 1.3–2.3) exposed vs unexposed. Higher in women OR 2.3 (95% CI, 1.5.3.5).
Wang 2021 (31)	China, contaminated area from Zn-Cd smelter	Dietary survey N = 467	CKD-EPI (GFR only)	Dietary survey estimated lifetime intake and divided individuals into high ≥2.2 g) and low (<2.2 g) intake. OR: 18.16 (95% CI, 1.75–188.85) high vs low
Thomas 2014 (32)	Sweden	Dietary survey Prospective cohort, no CKD at baseline.	National Hospital Discharge Register	Tertiles of estimated Cd intake. Cohort of Swedish Men (N = 40,378), Swedish Mammography Cohort (N = 33,929). No relationship between Cd intake and CKD using a Cox proportional-hazard model. HR in men: 0.97 (95% CI, 0.77–1.21), HR in women 0.74 (95% CI, 0.53–1.04).

Abbreviations: CI: confidence interval; CKD: chronic kidney disease; CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; DM: diabetes mellitus; EWAS: environmental wide association study; GFR: glomerular filtration rate; KNHANES: Korean National Health and Nutrition Examination Survey; NHANES: National Health and Nutrition Examination Survey; HTN: hypertension; OR: odds ratio; RBC: red blood cell.

3. Results

3.1. Entire study population

The baseline characteristics of the study sample are shown in Table 1 subdivided by the presence or absence of CKD ≥ stage 3. In the entire study population, the results of binary logistic regression are shown in Table 2. Cd increased the odds of CKD ≥ stage 3 by 1.89-fold, and lead (Pb) by 1.13-fold. Age over 50 years increased the odds of CKD ≥ stage 3 by 4.41-fold. Each decade a person lived in the smelter community increased the odds of CKD ≥ stage 3 by 1.34. Hypertension increased the chances of CKD ≥ stage 3 by 3.15-fold, and having T2DM increased likelihood by 2.51-fold. Smoking was protective from developing CKD ≥ stage 3 with an OR =0.26 (95% CI: 0.11–0.62, p < .002).

Table 1. Baseline characteristics for total sample and subgroups.

	CKD stage >3^a					CKD < 3					Total sample
	(n = 51)					(n = 766)					(n = 875)
Variables	M	SE	Lo	Hi	M	SE	Lo	Hi	p^b	M	SE	Lo	Hi
Age	63	1.9	24	86	45.9	0.5	19	86	.001	46.9	0.49	19	86
BMI	32.2	1	15	53	32.5	0.3	16	87	.0001	32.5	0.29	15	87
Arsenic (µg/L)	0.16	0.11	0	4	0.1	0.02	0	5	.6	0.1	0.02	0	5
Cadmium (µg/L)	0.24	0.09	0	3	0.08	0.05	0	4.5	.006	0.09	0.01	0	4.5
Lead (µg/dL)	3.92	0.7	0	33	2.16	0.08	0	23	.48	2.27	0.09	0	33
Mercury (µg/L)	0.06	0.04	0	2	0.06	0.01	0	7	.11	0.06	0.01	0	7
Duration of residence (years)	26.5	2.4	2	64	18.8	0.5	1	71	.11	1.93	0.04	0.1	7.1
	n	%			n	%			p^b	n	%
Females	28	54.9			510	61.9			.07	534	61.4
Smoking	9	17.6			256	31.1			.37	265	30.3
Age ≥50 years	42	82.4			310	37.6			.0001	352	59.8
GGT abnormal	18	16.5			106	12.9			.16	114	13
Smelter worker	9	17.6			46	21.8			.03	55	7.4
T2DM	16	31.4			93	11.3			.0004	109	12.5

^aProportion of individuals with CKD Stages ≤ 3 = 5.83% (95% CI: 0.045–0.076).

^bComparison of CKD ≤3 vs. >3 groups.

Table 2. Binary logistic regression of CKD Stage≥ 3 and covariates on Cd.

95% CI for OR
Independent variables	OR	Lower	Upper	p Value
Duration (decades)	1.34	1.10	1.64	.003
Male gender	1.32	0.62	2.80	.477
Abnormal GGT	1.06	0.43	2.60	.908
Hypertension	3.15	1.15	8.59	.025
Smelter worker	2.294	0.83	6.31	.108
Diabetes mellitus	2.51	1.24	5.07	.011
BMI	0.99	0.95	1.03	.512
Active smoker	0.26	0.11	0.62	.002
Cd	1.89	1.02	3.49	.044
Lead	1.13	1.02	1.24	.017
Mercury	0.63	0.22	1.78	.383
Age ≥50 years	4.41	1.98	9.82	.0001
<50 years old (n = 523)
Duration (decades)	1.38	0.79	2.39	.254
Male gender	1.98	0.40	9.86	.407
Abnormal GGT	0.10	0.004	2.68	.171
Hypertension	8.82	0.98	79.28	.052
Smelter worker	7.19	0.64	80.74	.110
Diabetes mellitus	4.38	0.53	36.13	.170
BMI	0.94	0.84	1.05	.253
Active smoker	0.46	0.09	2.49	.368
Cd	2.27	0.39	13.17	.362
Lead	1.02	0.76	1.37	.912
Mercury	–	–	–	^a
>50 years old (n = 352)
Duration (decades)	1.38	1.11	1.72	.004
Male gender	1.15	0.47	2.82	.764
Abnormal GGT	1.59	0.59	4.29	.359
Hypertension	2.46	0.77	7.85	.129
Smelter worker	2.22	0.67	7.41	.194
Diabetes mellitus	2.58	1.19	5.58	.016
BMI	1.00	0.95	1.05	.903
Active smoker	0.21	0.07	0.60	.003
Cd	2.13	1.01	4.47	.047
Lead	1.14	1.03	1.27	.013
Mercury	0.83	0.28	2.49	.738

^a16/523 values were non-zero, and the OR is inestimable.

3.2. Age <50 years

Among those less than 50 years old (Table 2), Cd was associated with a non-significant 2.27-fold increase in CKD ≥ stage 3. The effect of lead (Pb) was not significant and small (OR =1.02). Hypertension increased the odds of CKD ≥ stage 3 by 8.82-fold (p < .05).

3.3. Age ≥50 years

Individuals 50 years and older, Cd (Table 2) had a significantly increased odds of CKD ≥ stage 3 (OR =2.13, p < .05), as did Pb (OR =1.14, p < .01). Being resident in the community significantly (p < .004) increased the odds of CKD ≥ stage 3 by 1.38-fold for each decade of residence. T2DM had an increased risk of CKD ≥ stage 3 (OR =2.58, p < .02).

3.4. Meta-analysis

Meta-analysis of 11 studies (see Table 3), shown in Figure 1, selected for data quality and measures of CKD found an OR_FIXED of 1.20 (95% CI: 1.15–1.26, p < .0001). Several studies were excluded for data quality or measurement reporting issues. The fixed effects OR (OR_FIXED =1.20) was lower than the random effects (OR_{RANDOM =} 1.40; 95% CI: 1.19–1.65, p < .001) (Figure 1). OR_RANDOM is the correct estimate because the sample studies analysed were drawn from populations that differ from one another in ways that could affect the treatment effect and the analysis had high heterogeneity (ie different environmental sources and concentrations of Cd, ethnic differences, etc.). Q indicated the presence of high heterogeneity (Q = 109.6, p < .0001), and I² measures showed substantial heterogeneity (I² = 88.1%); tau (standard deviation) = 0.26 and tau² (variance) = 0.07 indicated a moderately high degree of true variance in the meta-analysis (0.26 and 0.07, respectively). Notably, heterogeneity is expected in meta-analysis, but extent of heterogeneity is important. In summary, Cd is associated with an increased odds for CKD ≥ stage 3 in the regression analysis of the Dallas smelter community (OR = 1.89, 95% CI: 1.02–3.49), which is not significantly higher than the meta-analysis of select published studies (OR = 1.20, 95% CI: 1.15–1.26) that reported individual Cd toxicology exposure levels.

Figure 1. Meta-analysis of cadmium and chronic kidney disease (CKD).

4. Discussion

Cd exposure increased the odds of CKD ≥ stage 3 by 1.89-fold. Covariates’ risk was increased by age >50 years (OR =4.41), decades lived in the smelter community (OR = 1.34), T2DM (OR = 2.15) and hypertension (OR = 3.15), smoking tobacco (OR = 0.26), and Pb exposure (OR = 1.13).

4.1. Age

In our study age ≥50 years increased risk of Cd-related CKD ≥ stage 3 by 4.41-fold, a common finding in other cross-sectional studies that examined the effect of age (Navas-Acien et al. 2009; Ferraro et al. 2010; Myong et al. 2012; Chung et al. 2014). In the OSCAR study evaluating the relationship between low level Cd exposure and early kidney damage the prevalence of proteinuria increased with age (Järup et al. 2000), and renal tubular damage occurred at lower total body Cd burdens than previously thought (urinary Cd concentration 1.0 nmol/mol creatinine). Cd exerts toxic effects on the kidney at low doses with no apparent threshold. A Cd risk assessment was carried out in relation to the background risk of CKD (Ginsberg 2012). Ginsberg used data from ten studies to determine the extra risk of CKD over baseline in women of two specific ages (47.8 and 75 years). At a daily Cd intake of 1.0 µg/kg/day (World Health Organisation tolerable intake level) in a 47.8 year-old woman there was a 25% increased risk of CKD. At a ten-fold lower intake (Agency for Toxic Substances and Disease Registry chronic minimum risk level) at age 47.8 years, there was a 2.7% increased risk and 3.4% at age 75. The Cd-induced effect was more pronounced with increasing age (10.5 cases per 1000 at age 75 vs. 1.5 cases per 1000 at age 30). The authors postulated that this was due to a decline in functional reserve and reduced homeostatic mechanisms. Taken together all of these studies show that age increases the risk of Cd-induced glomerular and tubular injury.

4.2. Diabetes mellitus

We found that diabetes mellitus increased the risk of Cd-associated CKD ≥ stage 3 by 2.51-fold. In other cross-sectional studies that examined this variable two studies detected an association (Lin et al. 2014; Kim et al. 2015) and two did not (Navas-Acien et al. 2009; Chung et al. 2014). The two most common causes of CKD are diabetes mellitus (DM) and hypertension. Others and we have shown an association between Cd exposure and T2DM (Little et al. 2020). Interestingly, not only does Cd potentiate diabetes-induced kidney injury but diabetics are also at increased risk for Cd-induced kidney damage (Buchet et al. 1990; Akesson et al. 2005). Cd exposure increased the risk of diabetic nephropathy (microalbuminuria) in Swedish women (Barregard et al. 2014), and Aboriginal and Torres Strait Islander Australians (Haswell-Elkins et al. 2008). Women may be more susceptible because they have higher total body Cd burdens for any given environmental exposure than men (Buchet et al. 1990), perhaps because like Pb oral Cd absorption increases with decreasing iron stores.

4.3. Hypertension

We also found that hypertension was associated with a 3.15-fold increased risk of Cd-associated CKD ≥ stage 3. In four other cross-sectional studies that examined the effect of hypertension, three found an association (Navas-Acien et al. 2009; Lin et al. 2014; Kim et al. 2015) and one did not (Chung et al. 2014). In an analysis of the Korean National Health and Nutrition Examination Survey (KNHANES) a two-fold increase in blood Cd concentration was associated with a small but statistically significant increase in systolic, 3.2 mmHg in women and 3.87 mmHg in men, and diastolic blood pressure, 2.24 mmHg in women and 1.98 mmHg in men (40). A doubling in serum [Cd] increased the risk of hypertension 31.5% in men and 18.6% in women with long-term Cd exposure. In a group of Thai women, urinary 20-HETE levels were increased in those with hypertension compared to normotensives suggesting a possible pathophysiologic mechanism (Kim and Lee 2012). 20-HETE is involved in renal salt balance and blood pressure regulation.

4.4. Smoking

Of the risk factors examined in cross-sectional studies smoking was the one with the most inconsistent results. We found that smoking was protective, as did one other report (Myong et al. 2012); others found either no effect (Lin et al. 2014), an effect in former but not current smokers (Navas-Acien et al. 2009), or an association in smokers (Chung et al. 2014). The reasons for these disparate results are unclear. Perhaps if information were available about former or present smoker the result would be different. It is also possible smokers have lower life expectancy, and are underrepresented in the older (≥ 50 years) age group, which is the age range of the preponderance of individuals with CKD ≤ Stage 3.

4.5. Meta analysis

In a prior systematic review of the association of Cd and CKD in workers and the general population, the authors concluded that a meta-analysis could not be conducted due to a lack of information about methods used, risk of bias and heterogeneity (Byber et al. 2017). Only one previously published meta-analysis has examined heavy metal exposure and CKD risk (Jalili et al. 2021). Cd exposure was associated with proteinuria (OR =1.25; 95% CI: 1.04–1.49, p = .02). A statistically significant association was not detected between Cd exposure and decreased eGFR (OR =1.09; 95% CI: 0.98–1.20, p = .10). Subgroup analysis showed that older adults (age >65 years), those with hypertension, and African Americans were at increased risk. Our meta-analysis of select published studies (Table 3) differs in that we only included studies that used a standard definition of CKD, based on the MDRD or CKD-EPI equations (Figure 1), and not different cut-off values (Hwangbo et al. 2011, p. 45), the Cockcroft Gault equation (Kaewnate et al. 2012) or percentiles of eGFR (Kim and Lee 2012). One of their included studies does not appear to include any data on eGFR (Wang et al. 2016). We included two additional recently reported studies (Lee et al. 2020; Tsai et al. 2021), for a total of 11 reports that contained sufficient information for a meta-analysis, with an estimated OR_FIXED of 1.20 (95% CI: 1.15–1.26, p < .0001). This indicates Cd exposure increases OR_FIXED of 1.20 (95% CI: 1.15–1.26, p < .0001) of CKD ≥ stage 3 by approximately 20%.

4.6. Possible mechanisms

Based on observations in both humans and animals it is biologically plausible that environmental or occupational Cd exposure could result in CKD. LMW proteinuria, an indicator of proximal tubular damage, is the most common Cd-related renal disorder in humans and is a good indicator of renal dysfunction in moderate to severe chronic Cd exposure. In a small series of patients from Japan, there was an inverse association between LMW proteinuria and creatinine clearance (Nogawa 1984). A similar association was also noted in solderers in Sweden (Järup et al. 1995). Tubular proteinuria increases with both higher urinary Cd concentration and increasing age (Mason et al. 1988; Shi et al. 2018). Chronic tubulointerstitial injury can result in a secondary decline in GFR. In addition, Cd exposure may also cause direct glomerular injury. When human glomerular endothelial cells are exposed to Cd in vivo, the permeability of the monolayer is increased (Li et al. 2016). This hyperpermeability is the result of the dislocation of adherens junction proteins β-catenin and VE-cadherin, as well as activation of the p38 MAPK (mitogen activated protein kinase) signalling pathway. When mesangial cells are exposed to Cd, cell death is initiated through apoptotic and apoptotic-like mechanisms that can be caspase dependent or independent (Li et al. 2016). Taken together these findings indicate that Cd exposure can result in both tubular and glomerular injury, which can act to reduce GFR.

5. Limitations

The major limitation of this study is the sample size of those with CKD ≥3, 51/857 (6%). The second limitation is the sample of convenience strategy. A third limitation is the inclusion of only African Americans, which ultimately strengthened the analysis by making the sample homogenous with respect to race. The novelty of the analysis is the extensive testing of study subjects included, and knowledge at the individual level of whether or not the participants worked at the smelters, how long and where they lived in the smelter community, and extensive medicinal chemistry evaluation in various test panels (CBC, heavy metals, lipid, renal, liver) for each individual. Our current data analysis includes follow-up data collected a decade after the Superfund Cleanup.

6. Environmental cadmium exposure and public health implications

Public health implications of environmental Cd exposure include medical, sociodemographic and economic factors. Obviously, the overarching public intervention is to remove the environmental sources of Cd exposure. In the present study, Pb screenings identified at risk individuals, targeted environmental interventions (eg Pb paint removal), but no sustainable reduction in blood Pb was achieved. The EPA Superfund project removed top soil and the City replaced public housing, which did reduce blood Pb levels. Cadmium screenings were incidental and part of the heavy metal panel.

In reference to specific health effects in the present study, renal function screening in Cd exposed populations should have allowed monitoring and possible interventions when CKD Stage 3B was reached, but Cd was not part of the initial Pb screenings in the 1970–1980s. Broadening indications for SGLT-2 inhibitor and GLP-1 families of antihyperglycemic drugs include therapy for CKD (Chen et al. 2022; Yau et al. 2022). These drugs have apparent potential for treating impaired renal function, but those who need treatment must be identified, hence the role of screening programs in areas at high risk of Cd exposure.

Sociodemographic and economic public health aspects of the present study in chronological order of the West Dallas smelter communities include: screenings for Pb in the early 1970s, as international warnings about Cd toxicity were emerging. The smelters in West Dallas were first established in the 1930s. In the 1940s–1950s, communities grew up around the smelters as employees and their families established neighbourhoods, before Pb and Cd toxicity were known. The City of Dallas built public housing in the area in the 1940s and 1950s to meet the need to housing. It took 20 years to launch an EPA Superfund cleanup after the first studies indicated deleterious effects from Cd and Pb exposure. Ultimately, Cd and Pb levels were decreased to safe levels.

The public health aspect of this anecdote is the sequence of events. Toxicity abatement and prevention in high-risk populations must consider the history of how neighbourhoods came to be sited in high-risk areas. The push-pull demographic factors drew people to the neighbourhoods for economic reasons (employment), and the City responded to housing demand 20 years before Pb and Cd toxicity were known. The EPA Superfund cleanup 40+ years later improved the safety profiles of public housing, elementary, middle and high schools that were sited within blocks of the smelters. In summary, among historically environmentally exposed Pb and Cd exposed populations, more rapid cleanups would prevent harm to many people. Health screenings should be conducted routinely to document high-risk areas and efficacious remediation.

7. Conclusions

In conclusion, Cd increases the risk of CKD ≥ stage 3 in a Dallas Superfund site lead smelter community. Meta-analysis of 11 studies supports the increased risk, and this association is biologically plausible. Collectively, the available data supports an increased risk of 30–89% of CKD ≥ stage 3 associated with Cd. Subgroups of patients may be particularly vulnerable such as the elderly, and those with diabetics and/or hypertensives.

Acknowledgements

Dr. Robert Reilly made huge contributions to this manuscript and should be the lead author. He made the decision to leave academia and requested that he not be an author. However, we acknowledge his contributions to this manuscript and the renal health of generations of US Veterans. We wish him all the best.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing is not available as this is protected by stipulations in the IRB.

Additional information

Funding

This study received no financial support.