Association Of Different Obese And Metabolic Phenotypes And The Risk Of Incident Type 2 Diabetes Mellitus:the Rural Chinese Cohort Study

As the main type of diabetes,type 2 diabetes mellitus(T2DM)has become an important public health problem worldwide.Although overweight/obesity can lead to a series of metabolic abnormalities that increase the risk of T2DM,different studies have found the existence of metabolically healthy overweight/obesity(MHO)and metabolically obese but normal weight(MONW)subgroups.We have limited knowledge of MHO and MONW prevalence and influence factors,the association between MHO/MONW and the risk of incident T2DM,and the interaction between obesity and metabolic risk factors and the effect on the progress of T2DM.Although MONW can significantly increase the risk of T2DM,the prevention of T2DM for normal weight people is not available.Therefore,we aimed to discuss the above problems on the basis of the established rural cohort population.Objectives1.To describe the baseline prevalence of MHO and MONW in rural Chinese population and factors affecting them.2.Body mass index(BMI)or waist-to-height ratio(WHtR)was used to define obesity.Participants who have none of metabolic disorders(hypertension,fasting glucose or triglyceride[TG],or high-density lipoprotein cholesterol[HDL-C]disorders)were considered metabolically healthy.We aimed to explore the association between BMI,WHtR,metabolic disorders,different obese and metabolic phenotypes and their interactions and the risk of incident T2DM.3.To develop risk prediction models for incident T2DM risk among normal weight rural Chinese people.Study design and MethodsData in part one was taken from a cross-sectional study conducted during 2007 to2008 of among 20194 individuals of Henan rural districts aged?18 years.After excluding underweight(n=715)and missing data of obesity and metabolic related variables(n=93),a total of 19386 participants were included in the analysis.The prevalence rates of MHO and MONW were standardized by the directed method using the the sixth national census in 2010 population(?18 years old).logistic regression was used to explore the influencing factors of MHO and MONW prevalence,estimating odds ratio(OR)and 95%confidence interval(CI).In part two,study participants were recruited from the Rural Chinese Cohort study,with a baseline examination from 2007 to 2008,and a follow-up study from2013 to 2014.The response rate was 85.5%,and the mean follow-up time was 6.0years.A total of 11862 participants without baseline underweight,diabetes,and without missing indicators of BMI and WHtR and biochemical data were included for analysis.The Cox proportional hazard regression model was used to calculate hazard ratio(HR)and 95%CI for the association between BMI,WHtR,metabolic disorders and different obese and metabolic phenotypes and risk of incident T2DM.Sensitivity analysis was conducted to assess the robustness of the above results.The interaction between BMI,WHtR and different metabolic disorders with the risk of incident T2DM was analyzed by multiplication model and additive model.In part three,we included data from 5706 people with normal BMI(18.5~23.9kg/m~2)without baseline T2DM in a rural Chinese cohort followed for a median of 6.0years.Univariate Cox regression analysis was used to screen potential predictors by estimating the ORs and their 95%CIs;predictors with P<0.05 were retained andevaluated further.Then stepwise model predictor selection(model 1)and leastabsolute shrinkage and selection operator(LASSO)(model 2)methods were used for multivariate Cox regression model selection.A 10-fold cross-validation was used to choose?and assess the predictive accuracy of the LASSO selection model.Discrimination(C statistic)and calibration(calibration plot)were used to assess the predictive accuracy of models 1 and 2.Internal consistency of the discrimination and calibration performance measures were evaluated by the Bootstrap technique.Tofavor the clinical implementation and workability of the risk model,we formulated the nomogram and Framingham Study risk score function models from the results of a better model.Results1.At baseline,the age-standardized prevalence of MHO and MONW was 4.60%and 37.07%,respectively.The age-standardized prevalence of MHO and MONW was decreased with age,despite of gender.Compared with metabolically abnormal overweight/obesity(MAO),being men,lower age,high physical activity,normal sleep time(6-8 hours),normal waist circumference and WHtR were independently associated with the healthy phenotype among the obese.Compared with metabolically healthy and normal weight(MHNW),factors associated with MONW phenotype in normal weight people were being women,higher age,low physical activity and abnormal waist circumference and WHtR.2.Compared with MHNW population,the risk of T2DM for MHO population was not statistically significant;regardless of obesity defined by BMI(adjusted HR1.74,95%CI:0.75-4.03)or WHtR(1.52,95%CI:0.67-3.47).Compared with MHNW population,participants with MONW can significantly increase the risk of incident T2DM,the aHR and 95%CI were 4.38(2.54-7.56)and 3.67(2.06-6.53),respectively.The risk of T2DM was increased for MAO participants,the aHR and 95%CI were9.48(5.56-16.16)and 9.01(5.18-15.68),respectively.The results of sensitivity analysis were consistent with above results.No matter the obesity was defined by BMI or WHtR,participants with baseline MHO had more favorable metabolism related indicators,fasting insulin level and insulin resistance during follow-up than or similar to participants with baseline MAO,but significantly worse than those with baseline MHNW phenotype.Moreover,the results of interaction analysis showed that T2DM risk was increased with the interaction of overweight/obesity and hypertension,overweight/obesity and abnormal triglyceride(TG),overweight/obesity and abnormal fasting plasma glucose(FPG),overweight/obesity and metabolically unhealthy for BMI criteria.As well,the T2DM risk was also increased with the interaction of obesity and hypertension,obesity and abnormal TG,obesity and abnormal FPG,obesity and abnormal high-density lipoprotein cholesterol,obesity and metabolically unhealthy for WHtR criteria.3.The results of prediction models of T2DM for normal weight people showed that 5 variables were selected(age,tea drinking,BMI,FPG and TG)in the model 1,3variables were selected in model 2(age,FPG and TG).The C statistic difference between model 1 and model 2 was not significant(model 1 vs.model 2:0.811 vs.0.803,P=0.062),and calibration plot showed good calibration for both model 1 and model 2.Internal validation demonstrated that adjusted discrimination C statistic and calibration slope for models 1 and 2 were similar with those from the models for the whole population,which indicated good internal consistency for our equations.Conclusions1.Compared with MAO,the healthy status of MHO was associated with young men,normal sleep time,high physical activity and without central obesity.Compared with MHNW,the unhealthy status of MONW was associated with old women,low physical activity and central obesity.The association of MONW and alcohol consumption warrants more research.2.Although the risk of incident T2DM for MHO population was not increased significantly compared with MHNW population,the metabolic related indicators of MHO were significantly worse than MHNW during 6-year follow-up.The T2DM risk was also increased with the interaction of obesity and metabolic disorders.Therefore,MHO may not a healthy status.Compared with MHNW,MONW significantly increased the risk of incident T2DM.3.The 3 variables included in model 2 can reach the prediction effect of model 1with 5 variables,so model 2 was superior to model 1.Among normal weightpopulation,age,BMI,FPG,TG and tea drinking all influenced the risk of incident T2DM,of which the main influencing factors were age,FPG and TG.