The Associations Between Exposures To Metal Mixtures And Changes In Fasting Plasma Glucose And Mediation Analysis: A Cohort Study Among Occupational Population

Diabetes is one of the most important chronic non-communicable diseases affecting human health,which imposes a serious burden on individuals and society.The number of people with diabetes in China has increased sharply in recent years,with 116 million people suffering from the disease in 2019,according to the International Diabetes Federation.Elevated fasting plasma glucose is an indicator of the incident and development of diabetes.Existing evidence suggests that chemicals in the environment,such as heavy metals,may play an important role in the development of diabetes.It is of great significance to investigate metal exposure and fasting plasma glucose changes for the diabetes prevention.Previous studies have examined the associations between metal exposures and diabetes,but the results have been inconsistent.It should be addressed that metal exposures from environment generally occur as a mixture or in different forms.The longitudinal study on multi-metal exposure and changes in fasting plasma glucose is scarce.Moreover,single-pollutant approach or simple two-way interaction used in previous studies have not been able to overcome the collinearity and correlation between multiple metal exposures,nor have they taken into account the independent effects of individual metal when multiple metals act simultaneously.Lasso regression analysis,one of the multi-pollutant approaches,is of unique value in dealing with the collinearity and strong correlation of mixed exposures.The mechanism of changes in glucose induced by metal exposure is unclear.Studies have reported that metal may cause changes in telomere length or mitochondrial DNA copy number,and changes in telomere length or mitochondrial DNA copy number may cause changes in fasting plasma glucose.However,limited population-based study avalible on the effect of telomere length or mitochondrial DNA copy number on the association between metal exposure and fasting plasma glucose.Metal smelting workers are the typical people exposed to metal mixtures.To explore occupationally multi-metal exposure and changes in fasting plasma glucose and relevant mechanisms,we carried out researches on the following three parts.In the first part of the study,based on the follow-up of manganese-exposed workers healthy cohort,we evaluated the association between 12 metal in blood cells,namely magnesium,manganese,iron,cobalt,copper,zinc,arsenic,selenium,rubidium,strontium,cadmium and lead,and changes in fasting plasma glucose.In the second part of the study,a mediation analysis was used to evaluate the mediating effect and mediating ratio of relative telomere length on the associations between metal exposure and changes in fasting plasma glucose.In the third part study,a mediation analysis was used to analyze and evaluate the mediating effect and mediating ratio of mitochondrial DNA copy number on the associations between metal exposures and changes in fasting plasma glucose.Our aim was to provide occupational population with scientific basis for diabetes.Part Ⅰ:The associations between exposure to metal mixtures and changes in fasting plasma glucose:a cohort study among occupational populationObjective:To investigate the associations between metal exposure in blood cell at baseline and changes in fasting plasma glucose based on a cohort study among occupational population.Methods:We conducted physical examinations and collected blood samples among workers in manganese-ferroy refiniry of Guangxi corparation in 2017（n=826）.By using questionnaire,we collected data on basic information,lifestyle habits,history of disease,medication,family history,occupational history and so on.702 of them（85%）were followed up in 2020 with 3 years follow-up time.We excluded individuals with diabetes,taking lipid-lowering drugs or lack of covariates at baseline.Finally,a total of 668 participants were included into the subsequent analysis.22 kinds of metal concentrations in blood cells were detected by inductively coupled plasma mass spectrometry.To assess the variability and reproducibility of metals in blood cells,320 subjects were tested for metal concentrations at baseline（2017）and follow-up（2020）,and 12metals in blood cells with internal correlation coefficients higher than 0.4showing good reproducibility were included for subsequent analysis.Fasting plasma glucose and lipids were detected by automatic biochemical analyzer.Changes in fasting plasma glucose was defined as the value of fasting plasma glucose at follow-up minus the value of fasting plasma glucose at baseline.In single-metal model analysis,we used the generalized linear regression model to explore the relationship between fasting plasma glucose changes and the 12 metals in blood cells as a continuous variable into the model in turn.Further,the subjects were divided into four groups according to the quantiles of the respective metals and the population with first quantiles of the metal concentration were treated as the reference group,adjusting for gender,age,work duration,BMI,occupation,smoking status,alcohol consumption status,CRP,dyslipidemia,family history of diabetes mellitus and hypertension.In the multi-metal model,we used the generalized linear regression model to explore the relationships between changes in fasting plasma glucose and the 12 metals in blood cells as a continuous or categorical variable into the model simultaneously,adjusting for covariates as in single metal model.Then we use lasso regression analysis to select the metal independently associated with the changes in fasting plasma glucose.Then restricted cubic spline regression analysis was used to explore the dose-response relationship between metal in blood cells and changes in fasting plasma glucose.To explore gender differences in the association between metal exposure in blood cells and changes in fasting plasma glucose,we re-performed the above analysis by gender stratification.In addition,we performed a stratified analysis based on smoking status in males and assessed the interation effect or joint effect of metal exposure in blood cells and smoking status on changes in fasting plasma glucose.Results:（1）Compared with 2017,fasting plasma glucose levels were significantly increased at follow-up（median,4.59 vs.5.44 mmol/L,P<0.05）.（2）The single-metal model analysis showed that each unit increase in log-10transformed levels of copper and zinc concentrations at baseline were associated with 2.429（95%CI:1.272,3.586）and 1.313（95%CI:0.377,2.250）mmol/L mmol/L changes in fasting plasma glucose in the overall participants（both P<0.05）.The categorical analysis showed significant changes in fasting plasma glucose increase for zinc and copper in blood cells while decrease for cobalt(all P_trend<0.05).In males,each unit increase in log-10 transformed levels of copper and cobalt concentrations at baseline were associated with 2.751（95%CI:1.304,4.197）mmol/L and-0.876（95%CI:-1.378,-0.373）mmol/L changes in fasting plasma glucose,respectively.The categorical analysis showed significant changes in fasting plasma glucose increase for copper in blood cells while decrease for cobalt(both P_trend<0.05).In females,each unit increase in log-10 transformed levels of copper and zinc concentrations at baseline were associated with 2.327（95%CI:0.512,4.143）mmol/L and 2.682（95%CI:1.113,4.250）mmol/L changes in fasting plasma glucose,respectively.The categorical analysis showed significant changes in fasting plasma glucose increase for zinc and copper in blood cells while decrease for magnesium(all P_trend<0.05).（3）The multi-metal model showed that each unit increase in log-10 transformed levels of copper concentrations at baseline were associated with 2.199（95%CI:0.716,3.682）mmol/L changes in fasting plasma glucose in the overall participants（P<0.05）.The categorical analysis showed significant changes in fasting plasma glucose increase for copper in blood cells while decrease for cobalt(both P_trend<0.05).In males,each unit increase in log-10 transformed levels of copper and cobalt concentrations at baseline were associated with 3.023（95%CI:1.124,4.921）mmol/L and-0.683（95%CI:-1.204,-0.162）mmol/L changes in fasting plasma glucose,respectively.The categorical analysis showed significant changes in fasting plasma glucose increase for copper in blood cells while decrease for cobalt(both P_trend<0.05).In females,each unit increase in log-10 transformed levels of zinc concentrations at baseline were associated with 2.368（95%CI:0.457,4.278）mmol/L changes in fasting plasma glucose（P<0.05）.The categorical analysis showed significant changes in fasting plasma glucose increase for zinc in blood cells(P_trend=0.057)and iron(P_trend=0.056)while decrease for magnesium(P_trend=0.003).（4）The multi-pollutant approach showed that lasso regression was used to select copper for overall,cobalt and copper for males and zinc for females;Restricted cubic spline regression analysis showed that in overall population,copper was linearly and positively associated with changes in fasting plasma glucose(P_overall<0.001,P _{non-linearity}=0.682);In males,copper was linearly and positively associated with changes in fasting plasma glucose(P_overall<0.001,P _{non-linearity}=0.550);and we observed a U shape dose-response relationship between cobalt in blood cells and changes in fasting plasma glucose(P_overall<0.001,P _{non-linearity}=0.013);In females,zinc was linearly and positively associated with changes in fasting blood glucose(P_overall=0.004,P _{non-linearity}=0.934);（5）In males,no significant interaction effect was observed for smoking and metal exposure on changes in fasting plasma glucose.There were joint effect between copper and smoking in changes in fasting plasma glucose.Compared to low copper-exposed non-smokers,the high copper-exposed（≥852.09μg/L）smokers with daily smoking number≥20 or pack-year≥20 had the highest changes in fasting plasma glucose with 0.594（0.305,0.883）and 0.542（0.207,0.878）,respectively.Compared with low cobalt-exposed non-smokers,smoking increased the change of fasting plasma glucose by(β（95%）0.221（0.024,0.417）mmol/L,and this effect was weakened by high cobalt level（≥0.12μg/L）withβ（95%）-0.045（-0.303,0.214）mmol/L.Conclusion:（1）In the overall participants,copper was linearly and positively associated with changes in fasting plasma glucose.Copper was linearly and positively associated with changes in fasting blood glucose,while cobalt was associated with changes fasting plasma glucose in a U-shaped dose-response relationship in males.In females,zinc was linearly and positively associated with changes in fasting plasma glucose.（2）There were joint effects between copper or cobalt and smoking in changes in fasting plasma glucose,but no interaction effects.The second part Mediation effect of relative telomere length on the associations between metal exposure and changes in fasting plasma glucose Objective:To evaluate mediation effect of relative telomere length on the associations between metal exposure in blood cells and changes in fasting plasma glucose.Methods:The participants were the same as in the first part.The relative telomere length of peripheral white blood cells at baseline was measured by real-time quantitative PCR.A generalized linear regression model was used to investigate the relationships between exposure to metals in blood cells and relative telomere length as well as relative telomere length and changes in fasting plasma glucose.A mediating model was used to evaluate the mediation effect and the mediation proportion of relative telomere length on the associations between metals in blood cells and changes in fasting plasma glucose.Results:（1）The median（P25,P75）for relative telomere length were22.28（16.89,30.35）in overall participants,21.92（16.70,29.21）in males and22.94（16.99,32.50）in females,respectively.（2）The association between metal in blood cells and relative telomere length:In overall participants,each unit increase in log-10 transformed levels of copper concentrations at baseline was associated with the decrease in relative telomere length by 0.904（1.556,0.252）.The categorical analysis showed significant relative telomere length decreased with the increase of blood cell copper level(P_trend<0.05).Further analysis showed that the significant inverse association between copper and relative telomere length in female,BMI≥25 kg/m²,smoking（no）,drinking（no）,dyslipidemia（no）,and no history of hypertension（P<0.05）.In males,we observed a positive correlation of relative telomere length(β（95%CI）of 0.129（0.012,0.245）with blood cell cobalt in the third percentile（P<0.05）.No significant association was found between copper in blood cells and relative telomere length（P>0.05）.In females,each unit increase in log-10 transformed levels of zinc concentrations at baseline was associated with the decrease in relative telomere length by 1.388（2.442,0.333）.The categorical analysis showed significant relative telomere length decreased with the increase of blood cell zinc level(P_trend<0.05).Further analysis showed that the significant inverse association between zinc and relative telomere length in BMI<25 kg/m²,dyslipidemia,no family history of diabetes and no history of hypertension（P<0.05）.Restricted cube spline regression analysis showed a negative and linear dose-response relationship between copper and relative telomere length(P_overall=0.011,P_nonlinearity=0.236).In males,there was an inverted U-shaped dose-response relationship between cobalt and relative telomere length(P_overall=0.070,P_nonlinearity=0.029).No significant dose-response relationship was found between copper and relative telomere length(P_overall=0.278,P_nonlinearity=0.326).In females,we observed a negative and linear dose-response relationship between zinc and relative telomere length(P_overall=0.069,P_nonlinearity=0.524).（3）The association between relative telomere length and changes in fasting plasma glucose:In overall participants,no significants associations were found between relative telomere length at baseline and changes in fasting plasma glucose.Further stratified analysis showed that the association between relative telomere length and changes in fasting glucose were significant in those who were females or no smoking（P<0.05）.In males,we did not observe a significant association between relative telomere length and changes in fasting plasma glucose（P>0.05）.In females,each unit increase in ln-transformed levels of relative telomere length at baseline were associated with the decrease in changes in fasting plasma glucose by 0.235（0.439,0.031）mmol/L.The categorical analysis showed significant decrease in change in fasting plasma glucose with the relative telomere length increased(P_trend<0.05).Further stratified analysis showed that the association between relative telomere length and changes in fasting glucose were significant in people those who having dyslipidemia and diabetes mellitus family history（P<0.05）.Restricted cubic spline regression analysis showed a significant negatively and linearly dose-response relationship between relative telomere length and changes in fasting plasma glucose in overall participants and females.However,we found no significant dose-response relationship between relative telomere length and changes in fasting glucose in males(P_overall=0.448,P _nonlinearity=0.257).（4）Interaction effects:In overall participants,we observed the multiplication model interactions of relative telomere length and copper in blood cells in changes in fasting plasma glucose(P_interaction=0.009).With the increase of the relative telomere length level（first to third interle）,theβ（95%CI）of fasting plasma glucose change were 0.258（-0.037,0.552）,0.316（0.089,0.544）and0.235（-0.003,0.472）,respectively.In females,we observed the multiplication model interactions of relative telomere length and zinc in changes in fasting plasma glucose(P_interaction=0.013).With the increase of the relative telomere length level（first to third interle）,theβ（95%CI）of fasting glucose change were0.313（-0.035,0.662）,0.203（-0.260,0.667）,and-0.202（-0.608,0.205）,respectively.In males,we observed the multiplication model interactions of relative telomere length and cobalt in blood cells in changes in fasting plasma glucose(P_interaction=0.070).No interaction between relative telomere length and blood cells copper on changes in fasting blood glucose was found.（5）Mediation analysis:In females,the mediation analysis showed that relative telomere length could mediate 11.5%of the association between zinc in blood cell and changes in fasting plasma glucose.Conclusions:In females,there was a negative and linear dose-response relationship between zinc in blood cell and relative telomere length,and a negative and linear dose-response relationship between relative telomere length and fasting blood glucose changes.Relative telomere length could mediate11.5%of the association between zinc in blood cell and changes in fasting plasma glucose.Part Ⅲ Mediation effect of mitochondrial DNA copy number on the association between metal exposure and fasting plasma glucose changesObjective:To evaluate the mediation effect of mitochondrial DNA copy number on the association between metal exposure in blood cells and changes in fasting plasma glucose.Methods:Based on the participants in the first part,we further eliminated individuals who have no data on mitochondrial DNA copy number.Finally,a total of 663 participants were included in the subsequent analysis.Using real-time quantitative polymerase chain reaction（PCR）to detect the peripheral blood leukocyte mitochondrial DNA copy number of the participants at baseline.A generalized linear regression model was used to investigate the relationships between exposure to metals in blood cells and mitochondrial DNA copy number as well as mitochondrial DNA copy number and changes in fasting plasma glucose.A mediation model was used to evaluate the mediation effect and the mediation proportion of mitochondrial DNA copy number on the associations between blood cell metals and changes in fasting plasma glucose.Results:（1）The median（P25,P75）for mitochondrial DNA copy number was45.57（37.27,54.57）in overall participants,44.40（37.10,53.17）in male and48.84（38.41,58.49）in female,respectively.（2）The association between metal in blood cells and mitochondrial DNA copy number:In overall participants,no significant association was found between copper and mitochondrial DNA copy number（P>0.05）.In males,each unit increase in log-10 transformed levels of cobalt and copper concentrations at baseline were associated with the increase in mitochondrial DNA copy number by 0.217 and 0.599,respectively（P<0.05）.However,no significant trend was found between cobalt or copper in blood cells and mitochondrial DNA copy number(P_trend>0.05).Further stratified analysis,the positive association between cobalt and mitochondrial DNA copy number were statistically significant for BMI<25kg/m²,non-smoker,drinking,dyslipidemia（yes）,family history of diabetes（no）,and no history of hypertension（P<0.05）.And the positive associations between copper and mitochondrial DNA copy number were statistically significant for BMI<25 kg/m²,smoking,drinking,family history of diabetes（yes）,and history of hypertension（yes）（P<0.05）.In females,each unit increase in log-10 transformed levels of zinc concentrations at baseline were associated with the decrease in mitochondrial DNA copy number by 1.073（95%CI:1.807,0.338）.The categorical analysis showed significant mitochondrial DNA copy number decreased with blood cell zinc level increased(P_trend=0.002).Further analysis showed that the significant inverse association between zinc and mitochondrial DNA copy number in BMI≥25kg/m²,dyslipidemia（yes or no）,no family history of diabetes,and no history of hypertension（P<0.05）.Restricted cube spline regression analysis showed a non-linear dose-response relationship between copper and mitochondrial DNA copy number in overall participants(P _overall=0.015,P _nonlinearity=0.021).In males,there was a positive and linear dose-response relationship between cobalt and mitochondrial DNA copy number(P_overall=0.062,P_nonlinearity=0.390).There was a positive and linear dose-response relationship between copper and mitochondrial DNA copy number(P_overall=0.032,P_nonlinearity=0.160).Restricted cube spline regression analysis showed a negative and linear dose-response relationship between zinc and mitochondrial DNA copy number in females(P _overall=0.010,P nonlinearity=0.253).（3）The regression analysis showed that there was no significant associations between mitochondrial DNA copy number and changes in fasting plsma glucose in overall participants or males or females（P>0.05）.Further,the obove results were confirmed by Restricted cube spline regression analysis(P _overall>0.05,P nonlinearity>0.05).（4）Interaction effects:In overall participants,mitochondrial DNA copy number and copper have a multiplication-model interaction(P_interaction=0.002)on the change of fasting blood glucose.With the increase of mitochondrial DNA copy number level（first to third tertiles）,theβ（95%CI）of fasting plasma glucose change were 0.384（0.19,0.577）mmol/L,0.124（-0.193,0.441）mmol/L and0.346（0.099,0.593）mmol/L,respectively.In males,no multiplicated-model interaction was found between mitochondrial DNA copy number and copper or cobalt on fasting plasma glucose changes(P_interaction>0.05).In females,mitochondrial DNA copy number and zinc have a multiplication-model interaction(P_interaction=0.080)on the change of fasting plasma glucose.With the increase of mitochondrial DNA copy number level（first to third tertiles）,theβ（95%CI）of fasting plasma glucose change were-0.057（-0.366,0.251）mmol/L,0.744（0.204,1.285）mmol/L and 0.389（0.021,0.758）mmol/L,respectively.（5）Mediation effects:we found no mediation effects of mitochondrial DNA copy number on the association between metal in blood cells and changes in fasting plasma glucose in overall participants or males or females.Conclusions:（1）In overall participants,there was a non-linear dose-response relationship between copper and mitochondrial DNA copy number,however,no significant dose-response relationship was found between mitochondrial DNA copy number and changes in fasting plasma glucose.There was a interaction effect of mitochondrial DNA copy number and copper on changes in fasting plasma glucose.（2）No interaction effect of mitochondrial DNA copy number and cobalt or copper in blood cells on changes in fasting plasma glucose in males.（3）In females,there was a negative and linear dose-response relationship between zinc and mitochondrial DNA copy number,however,no significant dose-response relationship was found between mitochondrial DNA copy number and changes in fasting plasma glucose.There was a interaction effect of mitochondrial DNA copy number and zinc on changes in fasting plasma glucose.（4）We found no mediation effects of mitochondrial DNA copy number on the association between metal in blood cells and changes in fasting plasma glucose in overall participants or males or females.