Association Of Metabolic Phenotypes Of Obesity With Incident Diabetes And Plasma Cytokine Levels In Different Phenotypes

The prevalence of overweight and obesity is increasing worldwide.Epidemiologic studies recognized high body mass index（BMI）as a major risk factor for diabetes.Notably,diabetes is one of the most prevalent non-communicable diseases worldwide.Lifestyle and other interventions targeting modifiable risk factors（e.g.overweight and obesity）are therefore imperative for the prevention of diabetes.The effects of obesity on diabetes are deemed to be mediated by adiposity-related metabolic abnormalities.However,recent studies have concentrated on a subset of individuals with obesity who have no or only a few abnormal metabolic features,named"metabolically healthy obesity"（MHO）,whilst those with unhealthy metabolic profiles are defined as"metabolically unhealthy obesity"（MUO）.In addition,there is a certain percentage of metabolic abnormal phenotypes in the normal weight population,known as"metabolically unhealthy normal weight"（MUNW）.Whether the MHO individuals are at higher risk of diabetes remains controversial.The inconsistency in previous studies may be due to a number of reasons,such as race,age,sex,sample size,follow-up time,and the definitions of MHO.Thus,longitudinal studies with large sample size and long-term follow-up are still needed.Furthermore,most previous studies have treated the MHO phenotypes as a static status.However,it remains unclear to what extent metabolic health and obesity status change over time and whether this change affects diabetes risk.Hence,the present study aims to explore the association between the transition of different obesity and metabolic status and diabetes incidence.The exact mechanism underlying the link between obesity and diabetes is not completely understood.The heterogeneous metabolic phenotypes associated with obesity（i.e.MHO and MUO）indicate that the association between adiposity and diabetes is not simply explained by absolute fat mass.Adipose tissue is the largest endocrine organ in humans and can secrete a large number of cytokines.Cytokines are considered to be a link between obesity and metabolic disorders such as insulin resistance（IR）and diabetes.However,it is unclear whether cytokines serve an important role in MHO and related metabolic disorders.Numerous studies have investigated the effects of obesity on cytokines,but they have not considered differences between obese individuals that develop metabolic disorders and those that remain relatively healthy.The heterogeneity of obesity may have contributed to inconsistencies in epidemiologic evidence.On the other hand,whether cytokines account for the metabolic differences observed between metabolically healthy and unhealthy individuals is relatively unknown.The study of metabolically healthy and unhealthy obese and normal weight individuals could provide a scientific basis for the interrelationships between inflammation,metabolic health,and obesity.However,little is known regarding the plasma cytokine signatures of these phenotypes and the conclusion remains controversial.Furthermore,most of these studies have focused on several or a small group of classical inflammatory cytokines,such as interleukin-1β（IL-1β）,IL-6,tumor necrosis factor-α（TNF-α）,adiponectin,and leptin.It is unknown how other cytokines（such as chemokines,growth factors）change and whether these variations are linked to metabolic health or obesity status.Therefore,it is necessary to systematically study a broad set of immune markers in different metabolic phenotypes.Based on the above background,the main purpose of this paper was as follows:Firstly,to investigate the associations of MHO and other obesity metabolic phenotypes and their changes over time with diabetes incidence.Secondly,to further investigate the associations of plasma cytokine levels with different obesity metabolic phenotypes.This study consists of two parts:Part I Associations of metabolic health,obesity phenotype and their change with incident diabetes:A prospective cohort studyObjectives:To investigate the associations of different obesity metabolic phenotypes and their 5-year changes with diabetes incidence.Methods:We examined 17309 participants derived from the Dongfeng-Tongji（DFTJ）cohort study who were free of diabetes,BMI<18.5 kg/m²,and missing values of BMI-metabolic status phenotype at baseline and were followed until Dec.31,2018.Participants were divided into 6 groups based on BMI（normal weight:18.5–23.9 kg/m²,overweight:24.0–27.9 kg/m²,or obese:≥28.0 kg/m²）and metabolic health status（healthy/unhealthy）defined by the Adult Treatment Panel III（ATP-III）criteria.Metabolic health was defined as meeting<2 of the 4 metabolic abnormalities（elevated blood pressure,hypertriglyceridemia,hyperglycemia,and low HDL-cholesterol）.The multivariable-adjusted hazard ratio（HR）and 95%confidence interval（CI）for incident diabetes were derived from the Cox proportional hazard regression model.Multivariate-adjusted models were adjusted for age,sex,smoking,alcohol drinking,physical activity,fruit consumption,vegetable consumption,family history of diabetes,and waist circumference.We performed various stratified analyses and sensitivity analyses to test the robustness of the results.After further excluding incident diabetes before Dec.31,2013,subjects with BMI<18.5 kg/m² and missing values of BMI-metabolic status phenotype at first follow-up in2013,we included 12206 subjects who were followed until Dec.31,2018,to study the associations of 5-year changes in BMI-metabolic health status with risk of incident diabetes.Participants were cross-classified by changes in BMI-metabolic health status during follow-up;overweight and obese categories were combined to ensure an appropriate sample size.Multivariate-adjusted models were adjusted for age,sex,family history of diabetes,and changes in lifestyle factors（smoking,alcohol drinking,physical activity,fruit consumption,vegetable consumption）and waist circumference between the baseline and the first follow-up.Results:During 152689 person-years of follow-up with a median follow-up of 9.9 years,we documented 2948 cases of diabetes.Compared with the metabolically healthy normal weight（MHNW）individuals,the multivariable-adjusted HRs（95%CI）were 1.46（1.26,1.71）for metabolically unhealthy overweigh（MUOW）and 1.68（1.33,2.12）for MHO individuals;however,the metabolically unhealthy individuals showed appreciably higher diabetes incidence（HR=2.22,95%CI:1.93,2.54 for MUNW,HR=3.15,95%CI:2.76,3.59 for MUOW and HR=3.81,95%CI:3.23,4.48 for MUO）.The results remained robust in sensitivity analyses.In stratified analysis,a significant interaction was found between BMI-metabolic health status and age（<65 years old,≥65 years old）on incident diabetes(P_interaction=0.02),but not between BMI-metabolic health status and sex or central obesity(P_interaction>0.05).Compared with MHNW individuals,MHOW and MHO individuals at age<65 years old were at increased risk for diabetes（all P<0.05）;however,a significant positive association was only found between MHOW individuals and diabetes（P<0.05）,but not MHO individuals among those aged≥65 years（P>0.05）.Among participants with metabolically healthy overweight or obesity（MHOO）at baseline,only 37.4%were unconverted and 44.3%converted to metabolically unhealthy overweight or obesity（MUOO）from baseline to the first follow-up.Compared with stable MHOO individuals,transitioning to MUOO from MHOO resulted in a multivariate-adjusted HR of 1.94（95%CI:1.44,2.62）for incident diabetes.However,even if participants with overweight/obesity maintained metabolic health,they were at an increased risk compared with stable MHNW individuals（HR=1.76,95%CI:1.26,2.45）.Irrespective of BMI,individuals with stable unhealthy metabolic status had higher diabetes risk than stable MHNW individuals（HR=3.42,95%CI:2.57,4.53 for stable MUNW and HR=5.21,95%CI:4.07,6.66 for stable MUOW）.On the other hand,compared with individuals with stable unhealthy metabolic status,transitioning to healthy metabolic status or normal weight phenotype was significantly associated with a lower risk of diabetes,the multivariable-adjusted HRs（95%CI）were 0.45（95%CI:0.34,0.60）for those who transited from the MUOO to the MHOO phenotype,0.57（95%CI:0.37,0.87）for those who transited from the MUOO to the MHNW phenotype and 0.68（95%CI:0.50,0.93）for those who transited from the MUOO to the MUNW phenotype;the multivariable-adjusted HRs（95%CI）was 0.37（95%CI:0.24,0.57）for those who transited from the MUNW to the MHNW phenotype.Conclusions:Firstly,even when metabolic health was maintained during 5 years,overweight/obesity remained a risk factor for diabetes.At the same time,the MHOO was a transient state and easily converted to an unhealthy phenotype over time,which was associated with an increased risk of diabetes.Secondly,individuals with unhealthy metabolism have the highest risk of diabetes in all BMI categories,and these individuals can reduce diabetes risk by improving their metabolic abnormality or obesity.In summary,obesity and metabolic abnormalities were risk factors of diabetes and highlight the importance of weight and metabolic risk factors management in individuals with overweight and obesity.However,mechanisms linking MHO to increased risk of diabetes remain incompletely understood,further study and in-depth research are needed.Part II Association of metabolic health,obesity phenotype and plasma cytokine levelsObjectives:To investigate the associations of plasma cytokine levels with different obesity metabolic phenotypes.Methods:We conducted a cross-sectional study in a total of 228 participants,including130 participants from the DFTJ cohort and 98 participants recruited from Shiyan Renmin Hospital（SYRH）.The diagnostic criteria for metabolic health in the SYRH population were consistent with the first part of our study.In the DFTJ population,metabolic health was defined as having none of 4 ATP-III metabolic abnormalities,and metabolic unhealthy was defined as having 3 or all of 4 ATP-III metabolic abnormalities to increase the statistical power.The Bio-Plex multiplex assay was used to measure plasma levels of54 cytokines,including interleukins（IL）,chemokines,growth factors,colony-stimulating factors（CSF）,interferon,tumor necrosis factor（TNF）,and adipokines.To compare the differences in plasma cytokine levels between the MHOO and MHNW individuals,and the differences between the different metabolic statuses in overweight/obese or normal weight individuals,independent sample t-test or Mann-Whitney U test was applied to test BMI-metabolic health status group differences for plasma cytokine levels in the two study populations.In each of the two metabolically healthy populations,the general linear regression model or unconditional logistic regression was applied to explore the relationship of BMI categories（BMI<24kg/m² or BMI≥24kg/m²）with the plasma levels of each cytokine（continuous or dichotomous dependent variables）,with adjustment for age,sex and use of anti-inflammatory drugs（i.e.nonsteroidal anti-inflammatory drugs or steroids）,and the fixed-effect meta-analysis was used to combine the results（Q test P>0.05;I²<50%）.Correlations between plasma cytokine levels and clinical-metabolic parameters were analyzed by Spearman rank correlation analysis in metabolically healthy people.Based on this analysis,unconditional logistic regression was applied to explore the relationship of plasma levels of each cytokine with metabolic health status（dichotomous dependent variables）,and these analyses were conducted separately in overweight/obese and normal weight individuals.Given that IR is closely linked to diabetes and can be used as a diagnostic criterion for metabolically unhealthy status,we performed sensitivity analysis based on cross-classification of BMI categories and IR（homeostatic model insulin resistance index[HOMA-IR]>2.5）.Finally,the mediation analysis was conducted to examine whether plasma cytokine levels mediated associations of BMI with metabolically unhealthy status（ATP-III criterion）and IR.Results:In both study populations,plasma levels of leptin and PAI-1 were significantly higher in all overweight/obesity subjects（MHOO or MUOO）than in MHNW subjects（all P<0.05）.MUOO subjects exhibited significantly higher plasma levels of leptin,PAI-1,IL-1β,IL-1ra,and IL-13 but lower levels of adiponectin than MHOO subjects（all P<0.05）.Meta-analysis showed that the MHOO status was significantly and positively associated with 3 pro-inflammatory cytokines,including 2 adipokines（leptin,PAI-1）and IL-6;and it was significantly and inversely with 4 cytokines,including anti-inflammatory adipokines（adiponectin）,homeostatic chemokines（CCL11,CCL27）and IL-17（all P<0.05）after adjusting for age,gender,and use of anti-inflammatory drugs.No statistically significant association was found for other cytokines（all P>0.05）.Among participants without IR,compared with the normal weight subjects,overweight/obese individuals had significantly higher plasma levels of leptin,PAI-1,adiponectin,CCL11,and CCL27（all P<0.05）.In addition,overweight/obese subjects without IR showed significantly and higher plasma visfatin level and lower plasma levels of anti-inflammatory cytokines（IL-4,β-NGF,and IL-12p40）（all P<0.05）.Meanwhile,within the metabolically healthy or non-insulin resistance population,these obesity-related cytokines were correlated with markers of glucose and lipid metabolisms.For example,plasma levels of leptin and PAI-1 exhibited a significant,positive correlation with HOMA-IR,and plasma adiponectin levels were significantly and negatively correlated with fasting plasma glucose and triglycerides while positively correlated with low-density lipoprotein cholesterol（r_sranging from 0.25 to 0.70,all P<0.05）.Among the overweight/obese participants,we identified 13 cytokines that were significantly positively associated with the metabolically unhealthy phenotype,including pro-inflammatory cytokines（visfatin,IL-1β,IL-18,and IL-2Rα）,inflammatory chemokines（CCL7,IL-8,and CXCL10）,colony-stimulating factors（GM-CSF）,anti-inflammatory cytokines（IL-4,IL-13,IL-1ra）and growth factors（LIF）,whereas the anti-inflammatory adipokine（adiponectin）was inversely associated with the metabolically unhealthy phenotype（all P<0.05）.Similarly,we found plasma levels of PAI-1,IL-13,IL-1ra,IL-4,and IL-18 were significantly and positively associated with IR（all P<0.05）.And 3 cytokines（leptin,β-NGF,and IL-12p40）levels were found to be positively associated with IR（all P<0.05）.No statistically significant association was found for the other cytokines（all P>0.05）.Among the normal weight participants,we identified 5 cytokines that were significantly and positively associated with the metabolically unhealthy phenotype,including pro-inflammatory cytokines（leptin,PAI-1,IL-1β）,colony-stimulating factors（GM-CSF）,and anti-inflammatory cytokines（IL-13）,whereas the anti-inflammatory adipokine（adiponectin）and chemokines（CCL2）levels were inversely associated with the metabolically unhealthy phenotype（all P<0.05）.Similarly,we found plasma levels of leptin,PAI-1,IL-13,and CCL2（inverse）were significantly associated with IR（all P<0.05）.No statistically significant association was found for the other cytokines（all P>0.05）.Mediation analysis showed that plasma levels of adiponectin and IL-13 mediated16.6%and 26.4%of the associations between BMI and metabolically unhealthy phenotype respectively（all P<0.05）.We also found that plasma levels of adiponectin,PAI-1,as well as leptin,mediated 19.0%,12.7%,and 54.1%of the associations between BMI and IR（all P<0.05）.Conclusions:Firstly,overweight/obese individuals with metabolically healthy or without IR all possessed abnormal levels of inflammatory markers,which is characterized by increased pro-inflammatory factors and decreased anti-inflammatory adipokines.Secondly,overweight/obese individuals with metabolically unhealthy or IR presented a higher pro-inflammatory profile,where the plasma level was commonly elevated.Thirdly,normal weight individuals with metabolically unhealthy or IR also presented a higher pro-inflammatory profile,suggesting that these cytokines are associated with metabolic abnormalities independently of obesity.Fourthly,plasma levels of adiponectin,leptin,PAI-1,and IL-13 contributed to part of the associations of BMI with metabolically unhealthy or IR.These findings provided scientific evidence and clues for the biological mechanisms linking obesity and related metabolic disorders.