Introduction

Chronic kidney disease (CKD), which has a high global prevalence and mortality (Bikbov et al. 2020), is a growing public health concern worldwide. China, which has the largest population and renal disease health care system, is facing an unprecedented economic and social burden of CKD (Liu 2013). In a nationwide survey conducted between 2009 and 2010, the prevalence of CKD in China has been reported to be 10.8% (Zhang et al. 2012). Risk factor identification coupled with early disease detection based on sensitive biomarkers has apparently become the core of CKD management.

Environmental factors were reported to account for 22% of the global burden of disease and 23% of deaths (Xu et al. 2018), and the hemofiltration function of human kidneys (almost 20% of the cardiac output of blood) makes kidneys especially vulnerable to environmental pollutants. Although some of the developing countries, including China, have achieved the rapid development of industrialization and urbanization during the past few decades, the accompanying environmental problems such as the release of metal pollutants, also posed some challenges to people’s health. Therefore, apart from traditionally known kidney health risk factors such as obesity, hypertension, and diabetes (Jha et al. 2013), increasing attention has been shifted to the associations between environmental pollutants represented by heavy metals and kidney health (Cui et al. 2018; Xu et al. 2018; Yang et al. 2019; Liu et al. 2020; Jalili et al. 2021).

Exposure to environmental metals, which can occur through air, food, drinking water, and soil (Soderland et al. 2010), is prevalent in daily life, especially in mineral-contaminated areas. Exposure to inorganic metal elements can adversely affect human health, and the kidney is one of the major target organs in metal exposure due to its characteristics of glomerular and tubular changes (Barbier et al. 2005; Orr and Bridges 2017). Renal tubular dysfunction is the most common manifestation of chronic cadmium and arsenic exposure, mainly through mechanisms including oxidative stress, DNA methylation, histone acetylation, impaired DNA repair, and apoptosis (Scott et al. 1993; Bork et al. 2010; Kim et al. 2015; Nair et al. 2015). The natural existence of heavy metals along with mining and smelting activities may increase both workers’ and nearby residents’ kidney damage risks, as metals accumulated in human kidneys are difficult to degrade (Soderland et al. 2010). People are naturally exposed to multiple metals, and synergistic or antagonistic effects caused by metal mixtures could be much more complex than previously thought (Braun et al. 2016; Shelley et al. 2012; Lin et al. 2014; Tsai et al. 2017). However, previous epidemiological studies have focused largely on the effects of exposure to a single or several metals on kidney health. Among these metals, major toxic metals, such as lead, cadmium, and arsenic, have received the most extensive attention (Ekong et al. 2006; Johri et al. 2010; Robles-Osorio et al. 2015). For other metals and trace elements, data are limited, and the effects of coexposure to multiple metals on kidney health are relatively unknown. In addition, although studies have previously explored the relationships between multiple metals and kidney function, in which the estimated glomerular filtration rate reflecting a severe decline in renal function was selected as the indicator, they may have ignored the broader population with early kidney damage and therefore underestimated the effects of metal elements on renal function (Yang et al. 2019; Liu et al. 2020). Considering that CKD is an irreversible disease, the identification of subclinical kidney damage (especially tubular damage) risk metals is of great preventive significance to patients with CKD.

Hunan province is located in the hilly areas of southern China, which is rich in mineral resources and dense population. The processing and refining of non-ferrous metals have caused serious heavy metal pollution to this region (Li et al. 2018; Du et al. 2019; Kan et al. 2021). In the current cross-sectional study, we analyzed the associations of coexposure to multiple plasma metals with local villagers’ renal tubular damage reflected by abnormal levels of retinol-binding protein (RBP) and β2-microglobulin (β2-MG). This is the first study of coexposure to 18 plasma metals and renal tubular damage in a Chinese population, and our findings will provide new clues and insights into the impact of exposure to metals on renal tubular damage.

Materials and methods

Study subjects

A total of 2764 villagers were recruited from August 2016 to July 2017 in three counties (Shimen, Qingshuitang, and Huayuan) of Hunan Province. The selected southern districts are heavy industrial bases and rich in non-ferrous metal resources. The local soil, air, and water are seriously polluted by heavy metals. Our inclusion criteria were as follows: (1) participants aged ≥ 18 years and (2) those who had been an area resident for at least 5 years. The exclusion criteria were as follows: (1) a lack of main data, such as plasma metal elements and biomarkers; (2) a lack of covariate data; (3) outliers with abnormal metal levels; and (4) serious diseases such as cancer and stroke. After screening, the number of participants ultimately included in this study was 1918. The study was approved by the Medical Ethics Committee of Xiangya Hospital, Central South University, and all participants signed informed consent forms.

Data collection and variable definitions

Each participant was required to complete a questionnaire. Trained investigators collected personal information, including age, sex, education level, and income level, as well as lifestyle information, including alcohol intake, smoking status, and exercise frequency. In addition, personal histories of diseases, including hypertension and hyperlipidemia, were also collected. In the questionnaire, education level was categorized as primary school or less, junior high school, high school, and college or more. Annual household income was classified as below ten thousand yuan, ten to thirty thousand yuan, thirty to fifty thousand yuan, and fifty thousand yuan or more. Drinking and smoking status were categorized as never, current, and past, while exercise in the last 6 months was categorized as yes or no. Clinical traits, including weight, height, systolic blood pressure, and diastolic blood pressure, were measured. Blood pressure measurements were taken in the sitting position after at least 5 min of rest. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. Hypertension was defined as a blood pressure ≥ 140/90 mmHg or self-reported physician-diagnosed hypertension. All blood samples were collected the next morning after 8 h of fasting and used to detect total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglyceride, and metal element concentrations. Hyperlipidemia was defined as a triglyceride level ≥ 2.26 mmol/L, a total cholesterol level ≥ 6.22 mmol/L, an HDL-C level < 1.04 mmol/L, an LDL-C level ≥ 4.14 mmol/L, or a previous physician diagnosis of hyperlipidemia. Morning urine was used to measure the levels of creatinine, RBP, and β2-MG. Since a β2-MG level ≥ 300 μg/g creatinine or an RBP level ≥ 300 μg/g creatinine was deemed abnormal (Aitio et al. 2007); subclinical tubular impairment was defined as a β2-MG level ≥ 300 μg/g creatinine or an RBP level ≥ 300 μg/g creatinine.

Measurement of exposure and biological indicators

The measurement of plasma metal element concentrations has been described in a previous article (Huang et al. 2021). Peripheral venous heparin blood specimens were collected after overnight fasting for at least 8 h. All samples were sent to the local laboratory for centrifugation within 2 h and were stored at − 80℃. Plasma concentrations of 18 metal elements, including vanadium (V), nickel (Ni), chromium (Cr), zinc (Zn), manganese (Mn), rubidium (Rb), iron (Fe), cobalt (Co), copper (Cu), selenium (Se), arsenic (As), molybdenum (Mo), tin (Sn), lead (Pb), antimony (Sb), cadmium (Cd), tungsten (W), and uranium (U), were detected by an inductively coupled plasma mass spectrometer (ICP–MS; Agilent Technologies, Waldbronn, Germany). Quality control of plasma metal measurements was managed by spiked recovery of pooled plasma samples (random sample extraction of 200 samples), three replicate measurements, and analysis of certified reference agents provided by ClinChek®controls (no. 8883 and no. 8884). Approximately 30 mL of the first morning midstream urine was collected from each participant and was kept frozen (− 20 °C) before the analysis. The concentrations of urinary RBP and β2-MG were determined at the hospital laboratory by immune transmission turbidimetry and immune latex turbidimetry, respectively. Considering that the concentrations of biomarkers in urine samples vary according to the water content in the urine sample, a standardized method of dividing the urinary tubular marker concentration by the urinary creatinine concentration was applied in our study.

Statistical analysis

Descriptive data on continuous variables are reported as the means ± standard deviations or the median (25th and 75th), and categorical variables are summarized as counts (percentages). Student’s t tests or Wilcoxon rank-sum tests were used for the comparison of continuous variables according to the skewness of the data, while categorical variables were compared using chi-square analysis. In addition, Wilcoxon rank-sum tests were also used to validate the distinction of metal element concentrations between participants with and without renal tubular damage. Concentrations of plasma metals and renal tubular damage indicators were log10-transformed in the following analyses because of their skewed distributions. Considering the moderate or high correlations among some plasma metals (Huang et al. 2021), LASSO penalized regression analysis (Lv et al. 2021) was performed first to select the plasma metals with the greatest effect on renal tubular damage risk. All the covariates and 18 metals were included in our LASSO regression model. To evaluate the relationship between the selected metal elements and renal tubular damage, we further applied single- and multimetal logistic regression models to calculate the odds ratios (ORs) and 95% confidence intervals (CIs). Multimetal linear regression analyses were used to assess the associations between the selected metal elements and renal tubular damage indicators. In addition, an RCS regression model was employed to analyze the dose–response associations of multiple-metal exposure and renal tubular damage risk. The overall association of coexposure to the selected metals with renal tubular damage indicators was calculated by BKMR analysis (Bobb et al. 2018). Finally, subgroup analyses were performed to evaluate whether the associations of the selected metals with renal tubular damage varied by different disease statuses (hypertension and dyslipidemia). All data were statistically analyzed using Stata software version 16.0 (Stata Corp, College Station, Texas, USA) and R 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). In all analyses, a two-sided value of P < 0.05 was considered statistically significant.

Results

Baseline characteristics and plasma metal levels of the study participants

A total of 1918 participants were recruited at three heavy industrial bases in Hunan Province, South China, from August 2016 to July 2017. The general characteristics of the participants are summarized in Table 1. The mean age of the participants was 55.5 years, and 60.0% of participants were women. Approximately 52.4% of the participants had hypertension, and 86.4% suffered from dyslipidemia. Among the total population, 56% of the participants had less than a primary school education, and approximately a quarter of the participants were smokers. The clinical values of urinary β2-MG and RBP were 78.5 (29.0 and 208.3) μg/g creatinine and 44.2 (20.7 and 197.8) μg/g creatinine, respectively. The concentrations of the 18 plasma metals are presented in Table 2. The levels of Cu, As, Se, Cd, and Sn were significantly higher (all P < 0.05) in the participants with renal tubular damage than in the participants without renal tubular damage.

Table 1 Baseline characteristics of the study participants (n = 1918)
Full size table
Table 2 Concentrations of metals in plasma among the study population with and without renal tubular damage (n = 1918)
Full size table

Selection of main plasma metals associated with renal tubular damage by the LASSO regression method

We conducted LASSO penalized regression models to reduce the dimensionality, and a total of 9 variables were selected within the optimal value of λ (the largest log (λ) with the minimum binomial deviance) based on tenfold cross-validation. Among the selected 9 variables, Se, Cd, As, and Fe were identified as the main metals associated with the risk of renal tubular damage (Fig. S.1).

Logistic regression analyses

Single-metal logistic regression analyses and multiple-metal logistic regression analyses were performed to evaluate the relationship of metal element concentrations with renal tubular damage (Table 3). In the covariate-adjusted single-metal model (Model 2), compared with those in the lowest quartile, participants in the highest quartile of As, Cd, and Se were at 1.46-fold (95% CI 1.09, 1.96), 1.53-fold (95% CI 1.14, 2.05), and 2.71-fold (95% CI 2.01, 3.67) higher risks for renal tubular damage, respectively. In the covariate-adjusted multimetal model (Model 3), compared with the first quartile, no significant associations were found between the fourth quartile of As or Cd and renal tubular damage, and the adjusted ORs (95% CIs) in the fourth quartile were 3.14 (2.27 and 4.34) for Se and 0.60 (0.44 and 0.83) for Fe.

Table 3 The associations between plasma metal levels and renal tubular damage (n = 1918)
Full size table

Dose–response relationships between the selected metals and renal tubular damage

In the multimetal adjusted RCS regression analyses, log-transformed plasma Fe (Poverall < 0.001, Pnon-linearity = 0.900), As (Poverall < 0.001, Pnon-linearity = 0.094), and Cd (Poverall < 0.001, Pnon-linearity = 0.066) showed linear associations with renal tubular damage risk. However, a significantly non-linear association was observed between log-transformed plasma Se and renal tubular damage risk (Poverall < 0.001, Pnon-linearity = 0.002), and the slope increased at the log10-transformed concentration of 1.79 μg/L (Fig. 1).

Fig. 1
figure 1

Multimetal-adjusted dose-response relationships of selected plasma metal levels with renal tubular damage risk. Models were adjusted for age, sex, education level, income level, drinking status, smoking status, exercise level, BMI, and the four metals (log-transformed). The black line and blue shadow represent the OR and 95% CI, respectively

Full size image

Linear regression analyses

To assess the associations of the selected metals with RBP and β2-MG, multiple-metal linear regression analyses were further performed (Table 4). The covariate-adjusted multimetal linear regression model (Model 3) showed a significant positive association of Se levels with the two indicators (β2-MG, β 0.61, 95% CI 0.43, 0.80, P < 0.001; RBP, β 0.67, 95% CI 0.46, 0.88, P < 0.001). Fe levels were found to be negatively associated with β2-MG (β − 0.32, 95% CI − 0.49, − 0.16, P < 0.001) and RBP levels (β − 0.29, 95% CI − 0.48, − 0.11, P = 0.002). However, no associations were observed between plasma As or Cd and renal tubular damage indicators in our model.

Table 4 The associations of plasma metals with β 2-MG and RBP levels (n = 1918)
Full size table

Bayesian kernel machine regression analyses

The BKMR model was first used to assess the overall effects of exposure to metal mixtures on RBP (Fig. 2A) or β2-MG levels (Fig. 3A). Compared to when the four metals were at the 50th percentile of concentration, a significantly positive relationship of the metal mixtures with RBP or β2-MG levels was observed when the four metals were above the 50th percentile of concentration, and the positive correlation increased with increasing metal mixture levels. In contrast, the metal mixture presented a negative relationship with RBP or β2-MG levels when the four metals were below the 50th percentile of concentration, although not all levels had statistically significant differences, and the negative correlation decreased with increasing metal mixture levels. When the metals were fixed at the 25th, 50th, and 75th percentiles of concentration, Fe was significantly negatively associated with RBP (Fig. 2B) or β2-MG levels (Fig. 3B) at the 25th, 50th, and 75th percentiles of concentration, while Se was significantly positively associated with the two indicators. In addition, Cd was significantly positively associated with only RBP levels, while the association of As with the two indicators showed no statistically significant difference. In the analyses of RBP, increased estimates of Fe, Cd, and Se were observed (from the 25th to 75th percentile of concentration) when the levels of metals were set from the 25th to the 75th percentile of concentration. However, the estimates of Fe increased and Se decreased (from the 25th to the 75th percentile of concentration) when the levels of metals were set at the 25th to the 75th percentiles of concentration in the analyses of β2-MG.

Fig. 2
figure 2

Joint effect of the metal mixture on RBP levels. Data were estimated by the BKMR model adjusted for age, sex, education level, income level, drinking status, smoking status, exercise level, and BMI. The concentrations of RBP and metals were log-transformed. A Overall effect of the metal mixture on RBP levels (estimates and 95% CI). The metal mixture at different percentiles of concentration was compared to the metal mixture at the 50th percentile of concentration. B Single-exposure effect of the individual metals on RBP levels (estimates and 95% CI). The individual metal effects on RBP levels at the 25th, 50th, and 75th percentiles of concentration when the metals were fixed at the 25th, 50th, or 75th percentiles of concentration

Full size image
Fig. 3
figure 3

Joint effect of the metal mixture on β2-MG levels. Data were estimated by the BKMR model adjusted for age, sex, education level, income level, drinking status, smoking status, exercise level, and BMI. The concentrations of β2-MG and metals were log-transformed. A Overall effect of the metal mixture on β2-MG levels (estimates and 95% CI). The metal mixture at different percentiles of concentration was compared to the metal mixture at the 50th percentile of concentration. B Single-exposure effect of the individual metal on β2-MG levels (estimates and 95% CI). The individual metal effects on β2-MG levels at the 25th, 50th, and 75th percentiles of concentration when the metals were fixed at the 25th, 50th, or 75th percentiles of concentration

Full size image

Subgroup analysis

The associations of Se and Fe with renal tubular damage were further assessed by multimetal-adjusted subgroup analyses according to different disease statuses (Table S.1). The significantly harmful effect of plasma Se on renal tubules was consistently observed in all subgroups and was not modified by disease status. However, in participants with hypertension or dyslipidemia, we found a significantly protective effect of plasma Fe on renal tubules (hypertension: adjusted OR for Q4 versus Q1 = 0.59, 95% CI 0.39, 0.89; dyslipidemia: adjusted OR for Q4 versus Q1 = 0.55, 95% CI 0.39, 0.78). However, among individuals without hypertension or dyslipidemia, no significant associations of plasma Fe with renal tubular damage were observed (Fig. S.2).

Discussion

It is well known that RBP and β2-MG in urine are sensitive biological indicators of renal tubular damage. Under normal conditions, low molecular weight proteins such as RBP and β2-MG can be effectively reabsorbed by proximal tubules after free filtration from the glomerulus. When metals cause injury to the epithelial cells of tubules, tubule reabsorption dysfunction causes these low molecular weight proteins to appear in urine (Prozialeck and Edwards 2010; Klotz et al. 2013). Although early tubular functional impairment is considered a benign condition, and its significance is often overlooked, previous studies have indicated that renal tubular damage caused by heavy metals is irreversible or not completely recovered and that subclinical tubular dysfunction may be associated with long-term renal function decline (Piscator 1984; Iwata et al. 1993; Liang et al. 2012). In the current study, we explored the associations of plasma As, Cd, Se, and Fe with renal tubular damage and the two sensitive biological indicators after dimensionality reduction by LASSO regression. Single-metal models showed that exposure to Se, Cd, and As was a risk factor for renal tubular damage, while multiple-metal models indicated that Se was positively associated, and Fe was negatively associated with renal tubular damage risk or the two sensitive biological indicators. In addition, RCS analysis revealed a non-linear relationship of selenium with renal tubular damage risk. BKMR analyses further verified the positive associations of metal mixtures with RBP and β2-MG levels when they were above the 50th percentile of concentration. Last, according to the subgroup analyses, the effects of Fe exposure on renal tubular damage presented a discrepancy between participants with and without hypertension or dyslipidemia.

Toxic heavy metals

Cd and As are widely known as toxic heavy metals, and their kidney damage effects have been extensively explored in previous epidemiological and animal experiments. According to the experimental findings, As and Cd can activate the caspase family and p-53 apoptotic pathway in proximal tubules, which leads to the overproduction of free radicals and inflammatory factors, further promoting oxidative stress and inflammation of renal tubular epithelial cells and ultimately causing pyroptosis or apoptosis of cells (Robles-Osorio et al. 2015; Zhang et al. 2021; Chou et al. 2019). Our single-metal models showed that plasma As and Cd were positively associated with renal tubular damage, which was in accordance with previous epidemiological studies in humans (Liang et al. 2012; Akesson et al. 2005; Hong et al. 2004; Nordberg et al. 2005). However, interestingly, the hazardous effects of traditional toxic metals on renal tubules in our study were modified by multimetal exposure. One possible explanation for this finding is that the hazardous effects of As or Cd exposure might be attenuated because of the complex interactions of other metals. For example, previous experiments indicated that the absorption of Cd was negatively associated with the concentration of Fe and that a sufficient Fe level could protect against the toxicity of Cd (Jamakala and Rani 2015). Notably, Fe was included in our multimetal models, and therefore, the current findings may inspire the exploration and understanding of traditional toxic metals.

Iron

Although no significant association between Fe and renal tubular damage was found in single-metal models, a protective effect of Fe on renal tubules was observed for the first time in our multimetal models, and they seemed to have greater effects on populations with hypertension and dyslipidemia. Fe, an essential element for human energy metabolism, oxygen transport and enzymatic activity, is mainly stored in hemoglobin and myoglobin. The renal tubular epithelial system has been identified as one of the important places for Fe handling. The maintenance of Fe homeostasis plays an important role in renal health because both Fe overload and Fe deficiency can negatively affect renal function and may lead to renal tubular epithelial cell injury (Van Swelm et al. 2020). Since the relationship of plasma ionic Fe with the actual iron load of the human body is relatively unknown, and there has been no previous population-based epidemiological research focused on Fe ion exposure and renal tubular damage risk, our results warrant further investigation.

Selenium

In our study, Se was significantly associated with renal tubular damage in both single-metal models and multimetal models. Se is one of the major antioxidant trace elements (Battin and Brumaghim 2009) due to its role as a cofactor of glutathione peroxidase. The primary pathway for Se exposure is food, followed by polluted water and air, and the toxicity of Se depends on its different chemical forms (Barceloux 1999; Ullah et al. 2019). The relationship between Se and the kidney has rarely been explored in epidemiological investigations. In a 2-year prospective study aimed at investigating the etiology of endemic nephropathy, the association of Se with the kidney was first reported in humans (Karmaus et al. 2008). A positive correlation was found between serum Se and renal biomarkers (including creatinine clearance and β2-MG levels) in adult offspring, which supported Se exposure as a risk factor for renal tubular injury. In addition, animal experiments have shown that the nephrotoxicity of Se varies greatly in different chemical forms (Nagy et al. 2015). Exposure to excess Se (usually much higher than normal human intake) may be harmful to the kidney (Eckhert et al. 1992; Johnson et al. 2000) and cause lesions of proximal curved tubules (Cukierski et al. 1989). Although excessive Se exposure is considered harmful, the general intake of Se in the population is often considered beneficial. Early studies suggested that Se intake cannot only ameliorate oxidative stress-related diseases, such as cardiovascular diseases, diabetes, cancer, and CKD but also moderate the toxicity of metals such as Cd and As, which has made it widely used as a dietary and medicinal supplement for a long time (Faure 2003; Zachara 2015; Zhang et al. 2016; Cai et al. 2016; Zwolak 2020). However, it is worth noting that an increasing number of recent studies have shown that the protective effects of Se may be uncertain or even harmful (Vinceti et al. 2017, 2018a). Observational epidemiological studies in seleniferous areas have provided the most powerful evidence for the assessment of Se toxicity, which found a positive correlation of environmental Se exposure with CKD risk factors such as blood pressure and blood glucose (Kuruppu et al. 2014; Li et al. 2017; Vinceti et al. 2018b, 2019). These studies suggested that we need to take Se species, exposure times, concentrations, and confounding factors into consideration, especially in cross-sectional studies based on various countries and population groups, when evaluating the health effects of Se exposure (Rayman and Stranges 2013). The lowest observable adverse effect level (LOAEL) for blood Se set by the EPA is 1350 μg/L (Lemire et al. 2012). Although the concentration of Se in our study was much lower than this standard, it should be noted that the upper safe limit of Se levels may be lower than previously thought (Vinceti et al. 2009). Moreover, the total Se levels measured in our participants’ plasma do not reflect the species of Se and therefore cannot fully assess its toxicity. Although available evidence that directly support our results is limited, we suspect that long-term environmental Se exposure may have adverse effects on the kidneys. However, more research is needed to verify our conclusions.

This is the first study of coexposure to 18 plasma metals and renal tubular damage in a Chinese population. The multimetal models and BKMR models used in our study helped to assess the association of multimetal coexposure with renal tubular damage, which better reflected the real exposure situation. Therefore, our findings will provide new clues and insights into the impact of exposure to metals on renal tubular damage. Nevertheless, several limitations remain in our study. First, this was a cross-sectional study and causal inferences cannot be made. There are no conclusions regarding whether early renal injuries in our study population are reversible and whether they can predict the occurrence of CKD in the future; thus, a prospective follow-up is needed to confirm our findings. Second, due to the substantial missing of data for some covariates during the questionnaire survey, potentially important covariates such as daily dietary intake, years of residence at the study site, family history, and family inheritance were not considered in our models and therefore may influence our results. Third, except for RBP and β2MG, renal tubular damage indicators such as Kim Injury Molecule 1 (KIM-1), alpha-1 microglobulin (α1-MG) and N-acetyl-β-D-glucosaminidase (NAG) were not included in this study. Therefore, the effects of exposure to heavy metals on renal tubular damage may be underestimated.

Conclusion

Our findings indicated that exposure to high levels of multimetal mixtures may increase the risk of renal tubular damage and that plasma Se and Fe were the main metals related to renal tubular damage when coexposure to multiple metals occurred. Higher levels of Se might increase the risk of renal tubular damage, while higher levels of Fe appeared to attenuate the risk of renal tubular damage. To further confirm these relationships and explore the specific mechanisms, large prospective studies coupled with animal experiments are needed.