Association Between Co-exposure To Multiple Metals And Metabolic Syndrome

With the acceleration of the industrialization process in China,the problem of heavy metal pollution caused by the long-term and frequent metal mining activities has become increasingly prominent,which was seriously threatening public health.To this end,the government issued the “Twelfth Five-Year Plan for the Comprehensive Prevention and Control of Heavy Metal Pollution”,in which lead(Pb),cadmium(Cd),chromium(Cr)and arsenic(As)were listed as key pollutants,considering copper(Cu),zinc(Zn)and other metals.Hunan was the key area for pollution control.However,the long-term health consequences of heavy metal pollution were unclear.Metabolic syndrome(Met S)is a recognized cause of cardiovascular disease and diabetes,and the five components of Met S include abdominal obesity,elevated blood pressure(BP),elevated fasting plasma glucose(FPG),elevated total triglycerides(TG),and decreased high density lipoprotein-cholesterol(HDLC)levels.Epidemiological evidences suggested that metals were involved in the development of Met S.Several studies have found that higher levels of As,Cd,and Pb were associated with increased risk of Met S,while Cu,Zn and Cr were negatively associated with Met S.However,opposite or negative associations have also been reported in other studies.Most of the existing studies mainly investigated the association of a certain single metal with Met S,and the conclusions were inconsistent,which might be due to the differences in the levels of metal exposure in subjects,the impact of co-exposure to multiple metals were not considered,and the biomarkers selected for each study were different.Regarding the issue above,the present study recruited three groups of participants with different typical exposure characteristics in Hunan Province as the subjects,and six metals including As,Cd,Pb,Cr,Zn,and Cu co-contaminated locally and emphasized by the“Twelfth Five-year plan” project were selected as the main pollutants.This study focused on analyzing the correlation between metals and different biomarkers under different exposure backgrounds and exploring possible factors affecting metal levels by simultaneously measuring the internal levels of metal exposure in plasma(e.g.p As,p Cd,p Pb,p Cr,p Zn,and p Cu)and urine(e.g.u As,u Cd,u Pb,u Cr,u Zn,and u Cu).On this basis,multivariate logistic regression and various supervised machine learning methods were further used to assess the association between co-exposure to multiple metals and the risk of Met S,and to explore nonlinear dose-response relationships and potential interactions.There are two main parts as follows:Part I The distribution,correlation,and influencing factors of plasma and urine metals levels in populations with different exposure characteristicsObjectives: To describe and compare the distribution of plasma and urine metal levels and their correlations in populations with different exposure characteristics,and further analyze the possible influencing factors of metal levels.Methods: Three groups of populations with different typical exposure characteristics(highPb-Zn-Cd exposure,high-As exposure,relatively low-exposure to heavy metals)in Hunan Province were selected as the studied subjects.The epidemiological questionnaire information and biosamples were collected,and the concentrations of six metals in plasma and urine were measured by inductively coupled plasma mass spectrometry.After exclusion of participants with invalid questionnaires and missing plasma and urine biosamples,a total of 3,337 subjects(1,192 with high-Pb-Zn-Cd exposure,1,092 with high-As exposure,and 1,053 with low-exposure)were finally included.The distribution of plasma,urine and creatinine-corrected urine metal(μg/g Cr)levels were described and compared among three groups by descriptive analysis methods and analysis of covariance(ANCOVA).The correlations between different metals and different biosamples(plasma and urine)were analyzed by Spearman’s rank test and partial correlation analysis.Furthermore,multiple linear regression was employed to evaluate the linear relationship between metals in plasma and urine.In addition,ANCOVA was used to construct multivariate regression model to compare the distribution of metal concentrations in subgroups stratified by age,gender,smoking,drinking,regular exercise,drinking water source and dietary intake frequency.Results: There were significant differences in the distribution of six metal levels in plasma and urine among the three groups with different exposure characteristics.Compared with the other two groups,the levels of Pb and Cd [geometric mean(GM): p Pb=2.38 μg/L,u Pb=10.16 μg/g Cr,and p Cd=0.21 μg/L,u Cd=4.44 μg/g Cr)] and other plasma metals in high-Pb-Zn-Cd exposed groups were significantly higher,but the urinary metal levels after creatinine correction and covariates adjustment were lower.The levels of u As(GM: 76.65 μg/g Cr)and u Cr(GM: 1.72 μg/g Cr)were higher,while the concentrations of Pb were lower in participants with high-As exposure.The p Zn level(GM: 805.99 μg/L)of the lowexposed residents was lower,but the u Zn level was higher(GM: 449.98 μg/g Cr).There was a moderate positive correlation between p Cu and p Zn in all groups,which was stronger after adjustment for gender,age and race,especially in low-exposed groups(Spearman rank correlation coefficient r=0.50,partial correlation coefficient r=0.55);p As,p Pb,and p Cd also showed moderate positive correlations(rs ranged from 0.39 to 0.45)in high-Pb-Zn-Cd exposed populations.The correlations between urinary metals were stronger than that of plasma metals.Except for u Cr,there were moderate to strong positive correlations between each pair of the other five urinary metals,especially in the lowexposed populations(rs ranged from 0.49 to 0.65).Among them,u Cu was closely related to other metals,the rs of Cu-Zn and Cu-Cd were greater than 0.60,even after creatinine correction(all rs>0.56).u As,u Pb,and u Cd were also positively correlated with each other at moderate intensity(rs ranged from 0.50 to 0.60 after creatinine correction).In contrast,the correlations between metals in different biosamples were weak(all rs≤0.40)and weakened in three groups with high-Pb-Zn-Cd exposure,high-As exposure,and lowexposure,among which the correlation between p As and u As was relatively strong(rs were 0.40,0.30 and 0.27),followed by Cd(rs were 0.37,0.22 and 0.15).However,after adjusting for confounders,the correlation between p As and u As was significantly weakened in highAs exposed populations.Under different exposure backgrounds,the levels of p As and u Zn were significantly higher in males than in females,while p Cu and u Cd levels were lower.With age,the levels of u Zn,u As,u Cd,and u Pb increased,while p Zn levels decreased,especially in the low-exposed group.The levels of Cd and Pb in smokers were significantly higher than in non-smokers,but lower in those who exercised regularly.In the population with high-Pb-Zn-Cd exposure,the levels of Zn and Cu were also higher in smokers and drinkers.Participants who drank unfiltered water had higher levels of p As and u Cd were,but lower levels of u As and u Pb.With the frequency of local rice intake increasing,the levels of Cd,As and Pb significantly increased.Under the background of high-As exposure,compared with the non-eating or low-frequency intake subgroups,the levels of u Cd and u Cr were higher in the populations with high-frequency intake of seafood,and the levels of u Pb and u As were higher in the subgroup of medium-frequency intake of preserved eggs,while less effect of seafood and preserved eggs consumption on the metal levels in the other two groups.Conclusions: There were strong correlations between Cu and Zn,and moderate positive correlations between six metals in urine.The pairwise correlations between plasma As,Cd and Pb and the correlations between the three and the corresponding metal levels in urine decreased sequentially in the populations with high-Pb-Zn-Cd exposure,high-As exposure,and low-exposure.The levels of metals in plasma and urine were affected by age,gender,smoking,drinking,exercise and diet.This study provided basic data for the analysis of biological effects of different metal biomarkers.Part II Association of co-exposure to multiple metals with metabolic syndrome in populations with different exposure characteristicsObjectives: To explore the associations of co-exposure to multiple metals with the risk of Met S under different exposure backgrounds,which provided a reliable basis for Met S prevention in populations with different metal exposure characteristics.Methods: Based on Part I,a total of 2,894 subjects(1,060 with high-Pb-Zn-Cd exposure,965 with high-As exposure,and 869 with low-exposure)were included after excluding those with missing basic information or important variables,cancer patients and renal dysfunction.Single-metal and multi-metal models were constructed by multivariate logistic regression to explore the associations of six metals in plasma and urine with the risk of Met S among populations with different exposure characteristics.Meanwhile,random effects meta-analysis was used to integrate the results of the three groups.Due to the moderately strong correlations between the six metals,supervised machine learning methods were further used to analyze the main effects: regularization methods [least absolute shrinkage and selection regression(LASSO)and elastic net(ENET)] were applied for variable selection,and the total joint effects of multi-metal mixture was evaluated by quantile-based g-computation(qgcomp).Besides,Bayesian kernel machine regression(BKMR)was used to explore nonlinear dose-response relationship and the interaction between metal pairs simultaneously,and restricted cubic spline was applied for validation.The above results were integrated to further explore the association between metal levels and Met S components.All models were adjusted for age,gender,race,smoking,drinking,education level,exercise,and e GFR,and urinary creatinine(adjustment for urine metal analysis).Results: Both single-metal and multi-metal models showed that p Zn levels were significantly positively associated with Met S in the population with low-exposure.Compared with the lowest tertile of p Zn levels(T1),the risk of Met S was increased by 67%(95%CI: 1.16,2.39)in the subgroup with higher levels of p Zn(T2);while the risk of Met S also increased in the highest subgroup(T3),but the odds ratio(OR)decreased [OR(95%CI)=1.44(1.00,2.08)](P-trend=0.018).However,no associations between plasma metals and Met S were found in the high-exposed populations.In the three groups with different metal exposure characteristics,u Cu and u Zn were significantly positively associated with the risk of Met S.With the levels of u Cu and u Zn increased(from T1 to T3),the risk of Met S also increased significantly(P-trend<0.05).In the multi-metal model,u Cd levels were significantly negatively associated with Met S,and for each interquartile range(IQR)increased in ln-transformed u Cd levels,the risks of Met S among three groups were reduced by 41%,45% and 31%,respectively.The above results were verified in the singlemetal effect estimation of BKMR.By integrating the results of the three groups through random-effects meta-analysis,we also observed that higher levels of u Cu and u Zn levels and lower u Cd levels were significantly associated with the increased risk of Met S in multimetal model,and the combined effects ORs and 95%CIs of u Cu,u Zn,and u Cd were 1.48(1.25,1.76),1.82(1.47,2.26),and 0.61(0.52,0.72),respectively.The nonlinear dose-response relationships fitted by BKMR were consistent with the tertile results,with an inverted "U-shaped" trend between p Zn levels and Met S risk in the lowexposed group,as verified by restricted cubic splines(P for nonlinearity was 0.002).Assessment of the overall effects of exposure to multi-metal mixtures on Met S by qgcomp showed that in the populations with high-Pb-Zn-Cd exposure and low-exposure,the risks of Met S increased by 33%(95%CI: 1.04,1.70;P=0.026)and 31%(95%CI: 1.06,1.62;P=0.012)for each quartile increased in ln-transformed levels of urinary metals mixture.In the concentrations range of metals significantly associated with Met S,higher levels of u Cd attenuated the Met S-induced effects of u Zn and u Cu,but no significant interaction between metal pairs on Met S was observed.Through the mutual validation and supplementation of LASSO and ENET regression,u Zn,u Cu and u Cd were screened out after ten-fold crossvalidation in the populations with high-As exposure and low-exposure,but no plasma metals associated with Met S were selected in the populations with different exposure characteristics.The weights and PIPs provided by qgcomp and BKMR also confirmed the dominant contributions of p Zn,u Zn,u Cu and u Cd to the positive and negative effects of Met S in the multi-metal models.Various statistical methods supplemented mutually and verified the robustness of the main effects.Further studies on the associations of metals with Met S components showed that higher p Zn levels were associated with the increased risk of abdominal obesity and elevated TG levels in the low-exposed group,with the ORs and 95%CIs of 1.31(1.09,1.58)and 1.57(1.30,1.89).Under different exposure backgrounds,higher levels of u Zn was associated with increased risks of abdominal obesity,elevated FPG and TG levels.For each IQR increased in ln-transformed concentrations of u Cu,the risk of elevated BP increased by 38%,45%,and 23%,respectively.Moreover,with the levels of u Zn and u Cu increased,the number of abnormal components of Met S showed an increased trend.u Cd level was negatively associated with abdominal obesity(P<0.01),with ORs and 95%CIs of 0.59(0.46,0.77),0.55(0.42,0.72)and 0.68(0.54,0.87),respectively.Conclusion: There were differences in the associations of plasma metals with the risk of Met S under different exposure backgrounds.Plasma Zn levels showed an inverted “Ushaped” positive non-linear association with Met S in the low-exposed populations,while no similar associations were found in high-exposed areas.The associations between urinary metals and Met S were relatively consistent.Various statistical models have demonstrated that increased levels of Zn and Cu or decreased levels of Cd in urine were associated with an increased risk of Met S.In the populations with high-Pb-Zn-Cd exposure and lowexposure characteristics,there were significant associations between co-exposure to six metal mixtures in urine and the risk of Met S,and the harmful joint effects were dominated by urinary Zn.This study provided a scientific basis for the prevention of Met S in populations with different metal exposure characteristics.However,due to the cross-sectional design,the findings need to be verified by prospective studies with larger sample sizes and further explored by in-vivo and in-vitro experiments in future.