Introduction

Diabetes mellitus (DM) is a serious threat to the global economy and health. International Diabetes Federation estimated approximately 463 million people with diabetes in 2019 and projected this number to reach 700 million by 2045(Federation, 2019). Metals play a key role in the physiological and biochemical processes occurring in the body. However, excess or deficiency of metals, especially essential trace elements, leads to bodily dysfunction. With rapid economic development and urban industrialization, heavy metal pollution has emerged serious concern. Therefore, our intake of metals through the respiratory tract, food, and drinking water is increasing day by day.

Recently, several studies have shown that metals, including antimony (Sb), arsenic (As), barium (Ba), cadmium (Cd), cesium (Cs), chromium (Cr), copper (Cu), manganese (Mn), molybdenum (Mo), palladium (Pd), selenium (Se), strontium (Sr), vanadium (V), and zinc (Zn) are associated with type 2 diabetes mellitus (T2DM)(Chen et al., 2020; Li et al., 2017; Xiao et al., 2018). A case–control study in Norway indicates that metals, such as Cd, Cr, nickel (Ni), iron (Fe), silver (Ag), and Zn, are significantly associated with DM (Hansen et al., 2017). Another study suggested that Cr and Fe are negatively associated with fasting plasma glucose (FPG), and Se is positively associated with FPG in men (Ge et al., 2021). A prospective cohort study established direct and indirect effects of dietary Mn on DM and hemoglobin A1c (HbA1c) (Du S et al. 2018). Various toxic metals such as arsenic, cadmium, and lead are known to induce oxidative stress in a variety of tissues and cell types (Ercal et al., 2001). A study in the Pakistani population shows that serum levels of antioxidant biomarkers, such as glutathione (GSH) and malondialdehyde (MDA), decrease due to arsenic exposure (Saba et al., 2020). Mercury (Hg) exposure adversely affects pancreatic beta cell development and function (Schumacher & Abbott, 2017). Low-dose mercury-induced oxidative stress and PI3K activation lead to Akt signaling-related pancreatic β-cell dysfunction (Chen et al., 2006). A cross section shows that urinary Cd, Cr, Mo, V, and Zn show a positive correlation with oxidative stress (Domingo-Relloso et al., 2019), an essential causative factor for the development of diabetes(Brownlee, 2005). Besides, strontium appears to be involved in anti-oxidation (Barneo-Caragol et al., 2018; Yalin et al., 2012) and adipose metabolism(Saidak et al., 2012). And exposure to high iron and glucose concentrations results in cellular oxidative damage and may initiate insulin secretory dysfunction in pancreatic β-cells by modulation of the exocytotic machinery (Blesia et al., 2021).

Several studies explored the relationship between single metals and diabetes, though their results differ (Ji et al., 2021; Soomro et al., 2019; Xiao et al., 2021). And the effects of multiple toxic metals on diabetes remain incompletely understood (Navas-Acien, 2019). These findings highlight the need to investigate the role of metal exposure in human risk of DM. China is a multi-ethnic country. In the past, research on the Chinese people mostly focused on the Han nationality. And the China Multi-Ethnic Cohort (CMEC) is a community population-based prospective observational study aiming to address the urgent need for understanding non-communicable diseases prevalence, risk factors, and associated conditions in resource-constrained settings for ethnic minorities in China. The Dong people in this study have a population of over one million and is the main ethnic minorities in Qiandongnan Prefecture, Guizhou Province. This population has fixed living habits and high genetic stability, and their data were more valuable than those of other floating population studies. Our objective was to evaluate the urinary exposure of trace elements (As, Cd, Cr, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sr, V, and Zn) in the Chinese Dong population and investigate the association of these trace elements with the risk of DM. We also aimed to evaluate how co-exposure to these trace elements may affect DM. The data from the CMEC study were used in the current study.

Method

Study population

The CMEC study, which has been reconsidered, is a population-based survey from July 2018 to December 2018, involving 5,792 Dong ethnic participants aged 30–79 years in Guizhou Province of China (Zhao et al., 2020). In brief, the sampling method used in the survey was multistage stratified cluster sampling. In the first stage, the Qiandongnan Prefecture, China were selected as our study sites. In the second stage, two communities (Liping County and Kaili City) in the settlement were selected by the local Centres for Disease Control and Prevention (CDCs), taking into account migration status, local health conditions and, most importantly, ethnic structure. In the final stage, all participants who met our inclusion criteria were invited to participate in our studies in consideration of both sex ratio and age ratio. In total, 1,313 participants were excluded. Exclusion criteria were the following:

(a) Those fasting for < 8 h (n = 425), (b) Patients with existing or previous tumors (n = 21), (c). Those lacking physical examination data, such as height, weight, etc. (n = 35), (d). Those without blood biochemical examination data of FPG, HbA1c (n = 16), (e). Those with a history of organ removal (n = 126), and (f). Those without any urine samples (n = 690).

Therefore, a cross-sectional study on 4,479 Dong people within this cohort was undertaken to investigate the association between urinary metal concentrations and diabetes. All survey participants signed prior informed consent forms. This study was approved by the Sichuan University Medical Ethical Review Board (K2016038) and the Research Ethics Committee of The Affiliated Hospital of Guizhou Medical University (2018[094]).

Assessment of covariates

A tablet computer with a self-developed application was employed to collect the information on socio-demographic characteristics, lifestyle behaviors, and personal health and medical histories. The education levels of the participants were categorized as junior high school or below, senior high school or vocational school, or college or beyond. Current smokers were defined as smoking > 100 cigarettes so far (Zhao et al., 2020). Former smokers were defined as quitting smoking > 6 months a year (Zhao et al., 2020). The drinkers were defined as drinking alcohol every week for ≥ 1 year. The height and weight were measured by trained medical professionals in accordance with standard procedures. A tea drinker was defined as a person drinking tea every week for a year or more. Body mass index (BMI) is equal to weight (kg) divided by the square of height (m2).

Assessment of blood biochemical indicators

The blood samples, collected at the baseline after fasting > 8 h, were transported to Guizhou KingMed Diagnostics Group Co., Ltd via cold chain logistics for same-day testing. The levels of FPG, glycated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL-C), and low-density lipoprotein (LDL-C) were determined using a fully automatic biochemical analyzer (HITACHI 7180, Tokyo, Japan).

Assessment of metals

The urine samples were stored at –20 °C before testing. Urinary metals/metalloids (As, Cd, Cr, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sr, V, and Zn) were measured by an inductively coupled plasma mass spectrometer (NexION 2000, PerkinElmer, Waltham, MA, USA). The standardized protocol, quality of control, and methods of accuracy evaluation have previously been described (Liu et al., 2021). The sample was placed at room temperature before dilution. Then, 1 mL original urine supernatant was diluted to 10 ml, with nitric acid (5%) (GR, Sinopharm Chemical Reagent Co., Ltd, Shanghai, China) and mixed well. Finally, the diluted sample was filtered through a 0.45 µm membrane and measured within 24 h. Internal standard solutions of Bi, Ge, In, 6Li, Sc, Tb, and Y at the concentration of 10 µg/L provided by PerkinElmer were used for internal online standardization, and the regression coefficient of the standard curve was not < 0.999. To check the accuracy, we used Seronorm™ Trace Elements Urine L-2 RUO (SERO, Billingstad, Norway). Percent recovery (i.e., amount of substance recovered on purification ÷ amount of substance originally taken) × 100) ranged from 82.92% – 115.71%. The limit of detection ranges from 0.0001- 0.1610 µg/L for the metals. The sample concentration below the limit of detection was expressed as half of the limit of quantification.

Diagnostic criteria

DM was defined as FPG ≥ 7.0 mmol/L or HbA1c ≥ 6.5% or self-reported history of DM (American Diabetes Association, 2018). Dyslipidemia was defined as TC ≥ 5.72 mmol/L or TG ≥ 1.70 mmol/L or LDL-C ≥ 3.40 mmol/L or HDL-C < 1.04 mmol/L or self-reported history of hyperlipidemia.

Statistical analysis

The continuous variables were expressed as median (interquartile range, IQR) because of skewed distribution; categorical variables were expressed as numbers (%). We evaluated the associations of 13 metals (As, Cd, Cr, Cu, Fe, Hg, Mn, Mo, Ni, Pb, Sr, V, and Zn) with diabetes through Logistic Regression models with adjustment for age, sex, education, smoking status, drinking status, tea consumption, BMI, dyslipidemia, and family history of diabetes. Principal component analysis (PCA) and weight quantile sum (WQS) regression were applied to investigate the effects of metal mixtures. The weights assigned to individual metals were estimated through PCA (weight range, 0–1 and the sum of all weights equal to 1) (Supplementary Fig. 1). WQS index was defined as under:

WQS index = As*0.11583 + Cd*0.08869 + Cr*0.11235 + Cu*0.04050 + Fe*0.08457 + Hg*0.01203 + Mn *0.04062 + Mo*0.08807 + Ni*0.07148 + Pb*0.06575 + Sr*0.10718 + V*0.08816 + Zn*0.08477.

Statistical analyses were performed with SPSS 25.0 (SPSS Inc, Chicago, IL) and EmpowerStats 2.0 (Inc. Boston, MA, USA). The inspection level was 0.05 (two-tailed).

Results

Demographic characteristics

Table 1 exhibits the baseline demographic characteristics of the participants in diabetic/non-diabetic groups. The total prevalence of diabetes was 9.5% (426/4479). The levels of FPG, HbA1c, TC, TG, LDL-C of the participants in the diabetic group were 3.13% –33.65% higher than those in the non-diabetic group (P < 0.05). No statistically significant differences were observed between diabetic and non-diabetic groups in education level and drinking status. Compared with non-diabetic participants, participants with diabetes had a higher percentage of men, older people, and tea drinkers. With respect to disease history, 300 (70.42%), 44 (10.33%) participants identified as dyslipidemia, and family history of diabetes in the diabetic group, respectively, whose ratio was higher than the non-diabetic group (P < 0.001).

Table 1 Demographic characteristics of the study participants
Full size table

Concentrations of urinary metals

Table 2 presents medians and interquartile ranges (IQRs) of metals in urine in the total population or by diabetes status. Only Hg and Ni were detected in < 80% of samples. Compared with the non-diabetes group, the concentrations of urinary Fe and Sr were about 11.84%/15.48% lower than those in the diabetes group, respectively (P < 0.05). Figure 1 depicts the correlations between metals. Most of the metals have a positive correlation, while only a few have a negative correlation (Mn-Cu, Pb-Cu, Mn-Fe, Mn-Ni, and Mn-V). The absolute value of the correlation coefficient ranged from 0.003–0.532. Moderately positive correlations were found for As-V, Cu-V, and Fe-Sr (P < 0.05).

Table 2 Concentrations of metals among the study population
Full size table
Fig. 1
figure 1

Heatmap illustration of partial correlation coefficient for metals in urine while adjusting for age, sex, BMI, education, smoking, drinking, tea consumption, dyslipidemia, and family history of diabetes

Full size image

Association of urinary metals with diabetes

Urinary Fe and Sr concentrations were negatively associated with diabetes, while other urinary metals/metalloids showed no association with diabetes in Model 1 (Table 3). In the adjusted model (Model 2), the associations of urinary Fe and Sr with diabetes remained significant (P trend < 0.05), and the ORs (95% CI) (the high vs. low) were 0.68(0.53–0.88) and 0.67(0.51–0.86), respectively, consistent with model 1. Figure 2 shows the smooth fitting curve between the urinary Fe and Sr concentration and the prevalence of diabetes. As the concentrations of urinary Fe and Sr increase, the risk of diabetes decreases (P < 0.05).

Table 3 Associations between metals and diabetes
Full size table
Fig. 2
figure 2

The smooth fitting curve of Fe and Sr in urine and the prevalence of diabetes. Curve-adjusted sex, ages, BMI, education, smoking, drinking and tea consumption, dyslipidemia, and family history of diabetes. The red curve in the middle represents the estimated value, and the blue curves on both sides represent the 95% confidence intervals

Full size image

Association of urinary metal mixture with diabetes

According to the scree plot and the criteria of eigenvalues of principal components > 1 (Supplementary Fig. 2), four principal components were selected, and the factor loading matrices of the four major PCAs are given in Supplementary Table 1. The three PC (PC 3) predominated by iron, nickel, and strontium, while the four PC (PC 4) were predominated by Mn and Pb. The PC 3 was negatively associated with diabetes and OR (95% CI) was 0.65(0.5–0.85). The PC 4 showed a positive association with diabetes and OR (95% CI) was 1.31(1.02–1.70). However, no association could be established between diabetes and the WQS index (Table 4).

Table 4 Associations between co-exposure metals and diabetes
Full size table

Discussion

In the current study, we explored the association of various metals with the risk of DM and evaluated how co-exposure to these trace elements may affect DM. We document that lower levels of urinary Fe and Sr were associated with an increased risk of DM during single metal exposure. While multiple metals are exposed together, the harmful effects primarily result from Mn and Pb, followed by Cr, and the beneficial effects mainly come from Fe and Sr.

It is suggested that Fe may affect glucose metabolism by affecting lipid metabolism.(Mayneris Perxachs et al., 2020) Increased hepatic iron load promotes fat regeneration, increased gluconeogenesis, and increased free glucose levels (Altamura et al., 2021). A research discovered that increased hepatic iron load is accompanied by insulin resistance (IR) (Abbate et al., 2021). Compared with the control group, the serum Fe level was higher in the diabetic group (Skalnaya et al., 2017). High Fe concentration increases the risk of DM (Hansen et al., 2017). A population-based cohort study indicated that serum ferritin levels, which represents the most commonly used indicators of iron storage, are higher in patients with T2DM, compared with non-T2DM (Diaz-Lopez et al., 2020). A cohort study from China demonstrated that higher serum ferritin was associated with a higher risk of DM (Chen et al., 2018). One cohort study in Japan showed that dietary Fe intake is a high-risk factor for DM (Eshak et al., 2018). In the current study, higher urinary Fe concentration is accompanied by a lower risk of DM, and the relationship between iron concentration and diabetes risk was U-shaped (Fig. 2). An in vitro study reported that increased levels of intracellular iron activate the Akt signaling pathway and 5' AMP-activated protein kinase (AMPK) pathways result in decreased liver glucose production(Varghese et al., 2018). In contrast, insulin-induced painful activation of AKT is attenuated because of the decrease in IRS1/2 protein levels(Varghese et al., 2018). Maybe this can help explain the U-shaped relationship between urinary Fe concentration and DM risk. Our results illustrate that urine Fe is a protective factor for DM, contrary to other studies. The concentrations of urinary metals vary from region to region. Because the study populations and regions are different, results are inconsistent.

A case–control study demonstrated that plasma Sr is inversely related with TC, LDL-C, and MDA and that as the plasma SR concentration increases, the risk of diabetes decreases (Chen et al., 2020). Additionally, an animal study showed that Sr has anti-diabetic effects by regulating the expression of related genes, lowering blood glucose levels, thereby improving resistance to insulin, leptin, and adiponectin (Maehira et al., 2011). Urinary Sr is negatively associated with the risk of diabetes in the present study, consistent with the results of the previous case–control study. However, a study indicated that compared with the first quantile, the second quantile shows that urinary Sr has a positive correlation with diabetes risk, but as the urinary Sr concentration continues to rise, the correlation is not statistically significant (Feng et al., 2015). This finding contradicted our research results. In this study (Feng et al., 2015), the median urinary Sr concentration was approximately 123.3 µg/L, but in the current study, the median concentration of urinary SR was about 62.2 µg/L. The urinary Sr concentration of the two studies is quite different, which may lead to inconsistent conclusions.

Although we found no association between Mn, Pb, Cr, and DM in single-metal-exposure analysis, PC4, dominated by Mn, Pb, and Cr, is positively associated with diabetes risk. Mn is an essential trace element. Mn is a component of manganese superoxidase (MnSOD) which protects mitochondrial DNA from glucose-induced oxidative damage to improve diabetic retinopathy (Sally A. Madsen-Bouterse and Kowluru 2010). An animal study reveals that Mn supplementation in normal mice can enhance the metalization and activity of MnSOD, and promote insulin secretion, thereby improving glucose tolerance (Lee et al., 2013). A cohort study in Japan found that high dietary manganese intake is associated with a reduced risk of T2DM in women (Eshak et al., 2021). Another study reported that high dietary manganese intake would reduce the risk of T2DM in women, mediated by interleukin-6 (IL-6) and high-sensitivity C-reactive protein (hs-CRP) (Jung Ho Gong 2020). Additionally, two cohort studies indicated the negative correlation between dietary manganese and glycosylated hemoglobin is statistically significant (Du S et al. 2018). However, a study revealed that blood Mn concentration has a J-shaped relationship with IR (Yang et al., 2020). And another cross-section study in China showed that blood Mn concentration has a U-shaped relationship with T2DM (Shan et al., 2016). The current study showed that as the urine Mn concentration increases, the risk of diabetes increases. A cross-section study revealed that diabetes patients have lower blood manganese and higher urine manganese compared with the control group (Kazi et al., 2008). This explains why our research shows that as the urinary Mn concentration increases, the risk of diabetes increases. A cohort study testified that higher Pb excretion is associated with a higher risk of DM (Wang et al., 2020). This is consistent with our findings. Pb can induce the occurrence and development of diabetes by promoting oxidative stress and altering intracellular signaling pathways, such as increasing resting intracellular Ca++. However, several studies approve that blood Pb has no association with DM, IR, beta-cell dysfunction, and FPG (Aziz et al., 2020; Feng et al., 2015; Moon, 2013). This may be because single metal exposure is inadequate to produce unhealthy outcomes, while multiple metal co-exposures increase health risks (Lv et al., 2021). The specific mechanism needs to be explored in future research.

This study has several strengths and limitations. First, we used the principal component analysis method to reduce the effect of collinearity between metals to a certain extent. Second, we used a generalized additive model to fit the dose–response relationship between metal concentration and diabetes risk. We also have some limitations. First, this study is a cross-sectional study that cannot directly explain the cause and effect. Second, this study did not involve eating habits that will affect the concentration of elements in the body. Although accurate measurement of dietary metals intake is challenging, it is interesting to explore the relationship between metals intake, plasma metals, urinary metals and diabetes. These will be explored in future research. Third, the diabetic patients in this study included newly-onset diabetic patients and those previously diagnosed with diabetes who may change their life behavior and dietary habits after the diagnosis of diabetes which may affect urine metal concentration.

Conclusion

The current study supports that co-exposure to multiple metals is associated with DM. The harmful effects are mainly from manganese and lead, and the protective effects are mainly from iron and strontium.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author.