Sex-Specific Associations of Urinary Metals with Renal Function: a Cross-sectional Study in China

Extensive studies have revealed the link between heavy metals and CKD. Compared to single meta-elements, mixture of metals reflect real-life metals exposure scenarios and are of interest. However, the mechanism of action of metal mixture on renal function is unclear. This study aimed to explore the potential relationship between urinary arsenic (As), cadmium (Cd), lead (Pb), manganese (Mn), and chromium (Cr) contents with estimated glomerular filtration rate (eGFR) levels in 2775 participants. The levels of metals in urine were determined by inductively coupled plasma–mass spectrometry. We used linear regression models and the Bayesian kernel machine regression (BKMR) to evaluate the association between metals and eGFR levels. In linear regression analysis, urinary As (β = 2.723, 95%CI: 0.29, 5.157) and Pb (β = 3.081, 95%CI: 1.725, 4.438) were positively associated with eGFR in the total population. In the BKMR model, a mixture of the five metals had a positive joint effect on eGFR levels, while Pb (PIP = 0.996) contributed the most to eGFR levels. Pb was positively associated with eGFR levels in the total participants and women. As was positively correlated with eGFR levels in women. Pb and eGFR levels were positively correlated when the other metals were set at 25th, 50th, and 75th percentiles. To the best of our knowledge, all five metals mixed exposure was positively associated with eGFR. Pb showed more important effects than the other four metals in the mixture, especially in women.


Introduction
Chronic kidney disease (CKD) is a public health problem of great concern. A cross-sectional study of CKD showed that the prevalence of CKD in China is high, especially in rural areas and specific geographical regions [1]. CKD imposes a huge burden on the health and economy of many countries and regions. The reported burden of CKD in China remains one of the leading causes of death, with 264 years of life lost per 100,000 population [2]. It increases the risk of end-stage renal disease, cardiovascular disease, and cognitive impairment [3], with high complications and mortality. Increasing evidence suggests an association between environmental compounds and chronic kidney function [4]. The general population is exposed to various environmental pollutants through drinking water or diet, which accumulate in the body over time and then develop health problems [5].
Yinxia Lin, Jiansheng Cai, and Qiumei Liu contributed equally to this work.
Heavy metal pollution is an important environmental issue because heavy metal ions are difficult to degrade. The kidney, as the main organ of excreting substances, is more vulnerable to heave metal contamination. eGFR is used to estimate the burden of kidney disease [6]. Several epidemiological investigations found association between arsenic (As), cadmium (Cd), lead (Pb), manganese (Mn), and chromium (Cr) and eGFR, suggesting that metal exposure impairs renal function [7][8][9][10]. Arsenic (As), cadmium (Cd) and lead (Pb) are known as nephrotoxic metals that increase the level of oxidative stress in the body, which plays a vital role in CKD [11,12]. Manganese (Mn) and chromium (Cr) are essential nutrients necessary for various biochemical and physiological functions, and deficiencies or excesses of these elements can have adverse effects on the body [13,14]. In addition, Mn and Cr can produce oxidative stress [14,15]. Pollutants in the environment are mixtures, and the public is exposed to multiple metals that coexist in the environment [16]. It is important to study the effects of multiple metals on renal function compared to single metals.
Bayesian kernel machine regression (BKMR) was used to estimate the health effects of multi-pollutant mixtures. This model can estimate the joint health effects of exposure to multiple risk factors, including exposure-response functions with nonlinear and nonadditive effects [17]. The model has been used by many scholars to assess the health effects of environmental mixed pollutant exposure [18,19]. In the present study, we used the BKMR model to analyze the relationship between metals and eGFR.

Study Population
A cross-sectional survey was conducted from 2018 to 2019 on the general population in the rural areas of Gongcheng Yao Autonomous County, which is located in Guangxi in the southwest region of China. Study subjects who met the following criteria were selected: (a) age ≥ 30; (b) residents of the study area. Four thousand three hundred fifty-six adults were recruited. Subjects were excluded according to the following criteria: (a) participants who did not complete the questionnaire or the questionnaire contained missing data on covariates (n = 429); (b) participants who had missing data on urinary metals detection (n = 851); (c) participants who had cancer or tumor (n = 118); (d) participants who had nephritis, renal deletion, renal insufficiency (n = 18); (c) eGFR < 60 mL/min/1.73 m 2 (n = 165). The remaining 2775 participants were analyzed (Fig. 1). Sample collection and questionnaire survey were conducted with participants who had given written informed consent. This study was approved by the Medical Ethics Committee of Guilin Medical University (No. 20180702-3).

Urinary Metal Determination
Midstream urine samples were collected and stored in the laboratory at -80 °C until metal detection was required. Concentrations of metals in urine: Pb, Cd, As, Cr, and Mn were measured using an inductively coupled plasma-mass spectrometer (ICP-MS, NEXION350; PerkinElmer, Inc., USA). Urine samples and diluted nitric acid were diluted at 1:9 and acidified overnight. Blank solutions of samples and reagents were determined under the operating conditions of the measurement standard series. Each sample was analyzed 3 times and the average value was taken. Internal quality control was performed using Seronorm™ Trace Elements Urine Level-1 and Level-2 (Sero AS, Billingstad, Norway). Two standard reagents were measured once in every 30 samples. The recoveries of the standard additions were controlled from 80 to 120%. The limits of detection (LODs) for Cr, Mn, As, Cd, and Pb were 0.0072, 0.0037, 0.0117, 0.0018, and 0.0009 μg/L, respectively. Samples with concentration below LOD were replaced by LOD /√2. The creatinine correction method was used to correct creatinine for the sample results at limit requirements.

Estimation of Kidney Function
Glomerular filtration rate (GFR) was estimated by age, sex, race, and serum creatinine levels. Studies showed the CKD-EPI creatinine equation is more accurate than the MDRD Study equation [20]. Therefore, the CKD-EPI formula was used to calculate eGFR as follows: where SCr is the serum creatinine concentration (in mg/dL) and age. When the subjects are men, κ = 0.9 and α = − 0.411; when the subjects are women, κ = 0.7 and α = − 0.329. Min indicates the minimum of SCr/κ or 1, and max indicates the maximum of SCr/κ or 1.

Data Collection and Covariates
A standardized and structured questionnaire was used to obtain the information on age, ethnicity (Yao or other ethnic groups), education level (≤ 6 years or > 6 years), smoking history (yes or no), and alcohol consumption (yes or no).
Smoking was defined as smoking at least one cigarette per day. Alcohol consumption was defined as drinking at least 50 g of alcohol or more once a month. Body mass index (BMI) was calculated by dividing weight by the squared of height (kg/m 2 ).

Statistical Analysis
Descriptive statistics were calculated for all demographic and clinical characteristics of the study population. Continuous variables are expressed as mean (SD) or median (IQR). Categorical variables were expressed as numbers and percentages. The Wilcoxon rank-sum test was used to analyze the distribution of urine metals. Pearson's correlation coefficients were calculated between urinary metals and eGFR. Metals content were log10 (lg) transformed to reduce their skewness.
Linear regression models were used to assess the relationship between the five metals as individual predictors and eGFR. Urinary metal concentrations were included in the model as quartile continuous variable. Covariates included sex, age, ethnicity, education, BMI, smoking, and alcohol consumption were selected and adjusted in the model as reported in a previous studies [1]. This model assessed the effects of co-exposure of five metals on renal function.
BKMR was used to estimate the overall association between metal co-exposure and eGFR levels, and this study considered possible interactions among five metals and nonlinear dose responses between five metals and eGFR levels [17]. We present the posterior inclusion probability (PIP) for each metal calculated by applying the BKMR model to identify the most important metal in the mixture. The PIP can be considered as a measure of variable importance, where higher values (closer to 1) show higher importance, and lower values (closer to 0) show lower importance.
Given the differences in serum creatinine concentrations between men and women, analyses were conducted separately by sex [6]. All data were analyzed using SPSS 20.0 software (SPSS Inc., Chicago, IL) and R (4.0.4 revised). All reported P values were made based on two-side tests with a significance level of 0.05.

Characteristics of Participants
A total of 2775 participants included 1018 men and 1757 women with a mean age of 56.76 ± 12.20 years and a mean eGFR of 93.57 ± 17.31 mL/min/1.73 m 2 (Table 1). Table 2 shows that urinary metal concentrations were higher in women than in men.

Correlation Analysis Between Metals
Pearson's correlation coefficients showed correlations between each pair of metals. Weak correlations were detected between Mn and eGFR (r = − 0.08), As and eGFR (r = − 0.06), and Cd and eGFR (r = − 0.08) in total population (P < 0.05). A positive correlation was observed among the five metals (r s values ranging from 0.16 to 0.45, P < 0.05). Similar results were observed in the sex group (Fig. S1).

Bayesian Kernel Machine Regression Analyses
The PIP values obtained from the BKMR model for each metal exposure are summarized in Table 4. The results showed that Pb had the highest PIP in the total population (PIP = 0.966) and women (PIP = 1.000). In the BKMR model, as shown in Fig. 2, we conducted the overall association analysis of the mixture. In the total population, we found a joint association of the metals mixture with eGFR levels. When all metals were above their 50th, the metals mixture showed a positive association with the eGFR levels, in contrast, a negative association was observed when metal concentrations were all below their 50th, as compared to that when all metals were at the 50th percentile (The 50th percentile of each metal was Cr 0.47 μg/g, creatinine, Mn 0.62 μg/g, creatinine, As 41.33 μg/g, creatinine, Cd 2.35 μg/g, creatinine, Pb 0.39 μg/g, creatinine). Similar results were observed in women.
We investigated potential nonlinear exposure-response relationships when the concentrations of other metals were kept at corresponding median concentrations. As shown in Fig. 3, urinary As and Pb correlate with eGFR in the total population and women. The single metal exposure-response relationship was consistent with the results of the single metals linear regression model.
As shown in Fig. 4, when the other five metals were set to p25, p50, or p75, a positive correlation between Pb and eGFR was observed in the total population (estimate: from 1.00 to 1.02) and woman population (estimate: from 0.967 to 0.997). However, no such association was found for the other four metals.
The bivariate exposure-response function is shown in Fig. S3. No interaction effect was found among the five metals.

Discussion
In this study, we comprehensively evaluated the relationship between urinary Cr, Mn, As, Cd, and Pb contents with eGFR levels and explored potential metal interactions among participants using the BKMR model. The BKMR study results are as follows: the mixture of five metals had a positive combined effect on eGFR levels, and Pb (PIP = 0.966) contributed the most to the eGFR levels. Furthermore, Pb and As are positively correlated with eGFR levels in women. Cr, Mn, As, Cd, and Pb showed no correlation with eGFR when analyzed as a mixture of contaminants rather than as a single contaminant. Compared with the metal levels reported in the other regions, for example, Wuhan (median values, 1.92 μg/L for Cr; 1.2 μg/L for Cd, 3.55 μg/L for Pb) [10] and Taiwan (median values, 0.1 μg/L for Cr; 1.7 μg/L for Mn, 78.9 μg/L for As; 0.8 μg/L for Cd; 0.85 μg/L for Pb) [21,22], the levels of Cd were higher in this region (median values, 0.27 μg/L for Cr; 0.34 μg/L for Mn; and 30.77 μg/L for As; 2.49 μg/L for Cd; 0.80 μg/L for Pb). A regional difference in human exposure to metals was suggested.
Pb can be toxic to the body even at low doses. Pb nephrotoxicity is manifested by proximal renal tubular nephropathy, glomerulosclerosis, interstitial fibrosis, and associated functional deficits. It can also enter the mitochondria of proximal tubular cells of the kidney, thereby impairing oxidative metabolism in the kidney [23]. The effects of Pb on the kidney have also been reported in many studies. Hui-Ju Tsai et al. found that high blood Pb concentrations were associated with the development of proteinuria and the reduction of eGFR [21]. A prospective study showed Fig. 2 Overall effect of the mixture estimates and 95% credible interval. A Total, B man, C woman. Data were estimated by Bayesian kernel machine regression, while adjusting for age, sex, BMI, smok-ing, drinking, education, and ethnicity. Univariate exposure-response functions and 95% credible intervals (shaded areas) for each metal with the other metals holding at the median that low concentrations of blood Pb can cause a decrease in eGFR and increasing the risk of CKD [24]. The results of a joint analysis study showed that blood Pb increased the risk of increased proteinuria and decreased eGFR [7]. The results of the association between urinary Pb and eGFR are inconsistent. A decrease in eGFR with increasing urinary Pb was found in a cross-sectional study conducted in Taiwan [22]. Xiao Chen et al. [25] found a positive association between urinary Pb and renal effect biomarkers such as urinary microalbuminuria, urinary N-acetyl-β-d-glucosamine, and urinary total protein, but no with eGFR. In the BKMR model, we found a positive correlation between urinary Pb and eGFR. A study in the USA found urinary Pb excretion increased with increasing eGFR [26]. Rufeng Jin et al. [27] also found that the rate of urinary metal excretion increased with increasing eGFR levels, where eGFR was more sensitive to Pb. Some studies suggest that this is poor hyperfiltration, which leads to subsequent renal adverse effects. In an animal experiment, GFR was positively correlated with blood Pb in rats for 1 month after Pb exposure and subsequently decreased [23,28]. Several studies have also regarded this as a reverse causality prediction [26,27,29]. Most metals are excreted through renal excretion and excretion is reduced after impaired renal function. Therefore, decreased eGFR and reduced metal lead to decreased metal levels in urine and increased metal levels in the blood.
Mn and Cr are essential trace elements, but their effects on the kidney have been less studied. In a Chinese crosssectional study, they analyzed the correlation between plasma metals and eGFR in the elderly population (over 90 years) and showed that dose-response relationship between Mn and eGFR was consistent with the trace element dose-effect curve, the U-type curve [30]. A Korean cross-sectional study found low blood Mn concentrations Fig. 3 Effects of single metal on eGFR. A Total, B man, C woman. Data were estimated by Bayesian Kernel machine regression, while adjusting for age, sex, BMI, smoking, drinking, education, and eth-nicity. Univariate exposure-response functions and 95% credible intervals (shaded areas) for each metal with the other metals holding at the median (1.28 μg/dL in participants with renal dysfunction) increase the risk and prevalence of renal dysfunction [31]. Jingli Yang et al. [32] also found a positive association between plasma Mn (median concentration, 9.34 μg/L) and urinary Mn (median concentration, 0.106 μg/g, creatinine) and eGFR. In this study, we found a positive correlation between urinary Mn and eGFR in women, with the urinary Mn levels of 0.78 μg/g, creatinine. Some studies of renal function in occupational Cr exposure have found negative or equivocal results [14]. In our study, results showed a positive association between urinary Cr and eGFR in women. Several studies found a negative correlation between urinary Cr and eGFR [10,22], but others did not found a correlation between urinary Cr and eGFR [23]. Due to the limitations of the cross-sectional study design, it is not possible to explain the causal relationship between Cr and kidneys, which requires more studies to confirm our findings. At present, whether Cr is a trace element necessary for humans is still controversial, but the effect of chromium-induced oxidative stress on the body should be worth noting. In the BKMR model, As was positively correlated with eGFR, and Cd was not associated with eGFR. It is commonly known that Cd and As are nephrotoxic heavy metals [11]. The results of some studies are consistent with our findings. In Taiwan, a study showed that a positive correlation between high urinary As and CKD [8]. Dongyue Wang et al. found no correlation between Cd in urine or blood and decreased eGFR [33]. Tsung-Lin Tsai et al. also found no association between urinary Cd and eGFR [22]. Many studies have shown that exposure to Cd or As exacerbates the development of kidney disease by producing toxicity through oxidative stress processes that generation of ROS, leading to lower glutathione levels, reduced superoxide dismutase and induction of DNA adducts [34][35][36][37][38][39][40].
Interestingly, we found that eGFR-associated metals were mostly different between men and women. First, urinary metal concentrations were higher in women than in men. This is due to the lower iron stores in women, which leads to the upregulation of the intestinal divalent metal transporter (DMT1), increasing the absorption of other metals [41]. We found that women's eGFR was more susceptible to metals, especially Pb. Studies have shown that metals affect metabolic disturbances in vivo or induce the upregulation of inflammatory biomarkers, leading to an increased risk of diseases such as cardiovascular disease, diabetes, and hypertension, especially in postmenopausal women. For example, estradiol (E2) disorder may be a risk factor for metabolic diseases in postmenopausal women, and E2 is susceptible to Pb interference [42]. The mean age of the women in this study was 55.87 years, so most were postmenopausal, and this population was more susceptible to metal effects. Furthermore, serum creatinine is limited by age and sex, so sex may also influence urinary metal excretion [6]. Although the current study cannot explain the sex-specific effects on the relationship between urinary metals and eGFR, we should focus more on the health effects associated with sex differences, especially the role of hormones in sex differences. Association (estimate and 95% credible intervals) of each metal increased from the 25th percentile to the 75th percentile with eGFR was observed when other metals in the mixture have been fixed at the 25th, 50th, and 75th percentiles. Estimate can be interpreted as the contribution of predictors to the response. A Total, B man, C woman. Data were estimated using the Bayesian kernel machine regression, while adjusting for age, sex, BMI, smoking, drinking, education, and ethnicity Our study has several advantages. First, BKMR analysis was used to analyze the single and combined effects of metals and to assess the effects of metal interactions on eGFR. Second, we performed different stratified analyses, which helped us to recognize that different metal exposure environments may affect the relationship between metal and eGFR. However, our study's limitations were noted. First, the present study was a cross-sectional study, and it was not possible to determine the causal relationship between metal exposure and eGFR. Second, certain factors may be affected by some measurements. For instance, it may affect creatinine through drugs and intestinal bacteria [43], which may exhibit extreme variation. Finally, we cannot exclude the possibility of false-positive results considering that our results were derived only from the excretion of these metals in the urine. Therefore, our findings need to be confirmed by further studies.

Conclusion
In the present study, urinary Cr, Mn, As, and Pb may be associated with eGFR. The mixture of five metals had a positive combined effect on eGFR levels, and eGFR is more susceptible to Pb. Sex-specific was also found in the association between heavy metals and renal function. Our findings provide a basis of reminding health researchers and government of the importance of environmental policy and legislative reform to improve human health. Future studies are needed to verify the causal relationships between heavy metals and CKD.