Association Of Plasma Lipidomics In Early Pregnancy With Gestational Diabetes Mellitus And The Potential Mediating Role Of The Lipid Metabolites

Gestational diabetes mellitus（GDM）is one of the common complications during pregnancy,which not only leads to short-and long-term adverse health consequences for both mothers and offsprings,but also brings a heavy economic burden to their families and society.GDM is an endocrine metabolic disease,characterized by insulin resistance,and lipid metabolism is considered to play an important role in insulin resistance.With the development of the high-throughput lipidomics technique,population-based studies have suggested that lipid metabolism disorder was closely related to GDM.However,the prospective studies were limited with relatively small sample sizes.Previous studies established the prediction models for GDM based on the conventional risk factors and these attained limited predictive powers;since most of the studies were performed among European-American population,the extrapolation of the results to Chinese populations remains to be verified.Thus,there is a need to identify the early biomarkers that could improve the prediction performance for GDM in Chinese pregnant women.In addition,lipid metabolism may play a potential mediating role in the associations between multiple endocrine and metabolic factors and GDM.Obesity and thyroid hormones are two important metabolic and endocrine factors during pregnancy,which are closely related to GDM,but the underlying mechanisms remain unclear.Mechanistic studies have suggested that obesity and thyroid dysfunction may affect lipid metabolism and initiate abnormal blood glucose afterwards,however,the evidence in pregnant women was scarce currently.In view of this,this study was based on the Tongji-Shuangliu Birth Cohort（TSBC）,which recruited a total of 7281 women aged 18～40 years with singleton pregnancy during2017～2020.A nested case-control study was established in this cohort,and targeted lipidomic measurement was performed on plasma samples in early pregnancy.The study aims to assess the associations of lipid metabolites with the risk of GDM,and to further explore the potential mediating roles of lipid metabolites in the associations of obesity and thyroid hormones with GDM.Part 1 Plasma lipidomics in early pregnancy and risk of gestational diabetes mellitusObjective:To evaluate the association of plasma lipidomics in early pregnancy with risk of GDM,and to assess their predictive capacity for GDM.Methods:The TSBC performed the lipidomics measurements in 2019.A total of 336women diagnosed with GDM through 75g oral glucose tolerance test（OGTT）at 24～28weeks between 2017～2019 were selected as cases,and each case was individually and randomly matched two healthy controls free of GDM on age（±3 years）and weeks of gestation at enrollment（±4 weeks）.Finally,a 1:2 matched nested case-control study was established.A total of 366 lipid metabolites were quantified in plasma in early pregnancy by ultrahigh-performance liquid chromatography-tandem mass spectrometry with targeted lipidomics techniques.The false discovery rate was applied to adjust the P values derived from the multiple hypothesis tests.Two sample t-tests and multivariable-adjusted conditional logistic regression models were used to identify the differential lipid metabolites showing statistical significance in at least one analysis,and least absolute shrinkage and selection operator（LASSO）regression with 10-fold cross-validation was further applied to select a set of lipid biomarkers that were associated with risk of GDM.The receiver operating characteristic curve and area under the curve（AUC）were used to assess the predictive capacity of the lipid biomarkers for GDM.Results:Two sample t-tests and multivariable-adjusted conditional logistic regression models identified 188 differential lipid metabolites.LASSO regression further identified 10 lipid metabolites that were significantly associated with GDM risk.After adjusting for confounding factors,phosphatidylinositol 40:6,alkylphosphatidylcholine36:1,phosphatidylethanolamine plasmalogen 38:6,diacylglyceride 18:0/18:1,and alkylphosphatidylethanolamine 40:5 were positively associated with risk of GDM,and the range of adjusted odds ratio（OR）per 1 log unit standard deviation（SD）increment in these lipids was 1.34～2.86,whereas sphingomyelin 34:1,dihexosyl ceramide 24:0,mono hexosyl ceramide 18:0,dihexosyl ceramide 24:1,and phosphatidylcholine 40:7 were negatively associated with risk of GDM,and the range of adjusted OR was 0.48～0.68.Addition of these 10 lipid metabolites to the GDM prediction model comprising traditional and clinical predictors resulted in a significant improvement in the AUC from0.70（95%CI[confidence interval,CI]:0.67,0.74）to 0.80（95%CI:0.77,0.83）.Conclusions:This study found a set of plasma lipid metabolites in early pregnancy that were significantly associated with risk of GDM.These identified lipid metabolites could significantly improve the prediction performance for GDM beyond the traditional and clinical predictors.Part 2 Associations of obesity indicators with gestational diabetes mellitus and the mediating role of the lipid metabolitesObjective:To evaluate the associations between obesity indicators in early pregnancy and risk of GDM.To identify the obesity indicators-related lipid metabolites and to explore the mediation effects of obesity indicators on GDM through related lipids.Methods:A total of 7013 participants from the cohort with data on physical examination were selected in this part,and multivariable-adjusted logistic regression models were used to evaluate the associations between obesity indicators in early pregnancy and risk of GDM.Participants in the nested case-control study（336 cases and672 controls）in Part 1 were selected for the mediation analyses.First,multiple linear regression models were used to identify the obesity indicators-related lipid metabolites;second,based on the significant lipid metabolites,LASSO regression was applied to select and create the obesity indicators-related lipid scores;third,multivariable-adjusted conditional logistic regression models were used to evaluate the associations between lipid scores,individual lipid metabolites and risk of GDM;finally,mediation analyses were performed to explore the mediation effects of maternal obesity indicators on GDM through the lipid metabolites.Results:Among 7013 participants,after adjusting for confounding factors,body mass index and waist in early pregnancy were positively and linearly associated with risk of GDM,and each 1-SD increase in body mass index and waist（i.e.,3.0 kg/m² and 8.0 cm）were associated with 35%（95%CI:23%,48%）and 29%（95%CI:18%,41%）increased risk of GDM,respectively.Among the nested case-control study with 1008 women,multiple linear regression models and LASSO regressions selected 63 and 39 lipid metabolites,and created the lipid scores for body mass index and waist,respectively.Two lipid scores were positively and linearly associated with risk of GDM,and the ORs of GDM were 1.57（95%CI:1.25,1.96）and 1.44（95%CI:1.17,1.78）for each 1-SD increase in body mass index and waist-related lipid scores,respectively.Mediation analyses indicated body mass index and waist-related lipid scores mediated 83.4%and84.2%of the associations between body mass index and waist in early pregnancy and GDM risk,respectively.Further analyses revealed that 13 lipid metabolites（including 6glycerophospholipids,5 glycerolipids and 2 sphingolipids）positively mediated the association between body mass index and GDM risk,and the mediated proportions were ranged from 10.6%to 24.9%;10 lipid metabolites（including 5 glycerophospholipids,4glycerolipids and 1 sphingolipid）positively mediated the association between waist and GDM risk,and the mediated proportions were ranged from 18.3%to 45.6%.Conclusions:Higher body mass index and waist in early pregnancy were associated with the plasma lipidomic profiling,and may increase GDM risk by disturbing the glycerophospholipids and glycerolipids metabolisms.Part 3 Associations of thyroid hormones with gestational diabetes mellitus and the mediating role of the lipid metabolitesObjective:To evaluate the associations between thyroid hormones in early pregnancy and risk of GDM.To identify the thyroid hormones-related lipid metabolites and to explore the mediation effects of thyroid hormones on GDM through related lipids.Methods:A total of 6015 participants from the cohort with complete measurements on serum thyroid hormones（including f T3[free triiodothyronine],f T4[free thyroxine],and TSH[thyroid-stimulating hormone]）in early pregnancy were selected in this part,and multivariable-adjusted logistic regression models were used to investigate the associations between f T3,f T4,TSH and f T3/f T4 ratio and risk of GDM.Based on the nested case-control study in Part 1,a total of 883 participants（298 GDM cases and 585 controls）with complete data on thyroid hormones and plasma lipid metabolites were selected for the mediation analyses.Thyroid markers that associated with GDM risk were analyzed.Multiple linear regression models were firstly used to identify the thyroid markers-related lipid metabolites;second,based on the significant lipid metabolites,LASSO regression was applied to select and create the thyroid markers-related lipid scores;third,multivariable-adjusted logistic regression models were used to evaluate the associations between lipid scores,individual lipid metabolites and risk of GDM;finally,mediation analyses were performed to explore the mediation effects of thyroid markers on GDM through the lipid metabolites.Results:Among 6015 participants included in this part,after adjusting for confounding factors,f T4 levels in early pregnancy were significantly associated with GDM risk in a non-linear manner(P_overall=0.016,P_nonlinearity=0.007,reverse J-shaped),and the inverse association was observed when the concentrations were<1 ng/d L and the positive association was observed when>1 ng/d L;f T3/f T4 ratio was positively and linearly associated with risk of GDM(P_overall=0.033,P_nonlinearity=0.657),and each 1-log unit increase in f T3/f T4 ratio was associated with the 2.09-fold（95%CI:1.20,3.65）increased risk of GDM;no significant associations were observed between f T3,TSH and the risk of GDM.Among the 883 women within the nested case-control study,multiple linear regression models and LASSO regressions selected 12 and 18 lipids,and created the lipid scores for f T4 and f T3/f T4 ratio,respectively.The lipid score for f T4 was neither linearly nor non-linearly associated with GDM risk,thus the mediation analysis was not further performed;while the lipid score for f T3/f T4 ratio was positively and linearly associated with risk of GDM,and the OR of GDM was 1.55（95%:1.28,1.87）for each1-SD increase in score.Mediation analysis indicated f T3/f T4 ratio-related lipid score mediated 57.0%of the associations between f T3/f T4 ratio in early pregnancy and GDM.Further analyses revealed that 6 lipid metabolites（4 glycerophospholipids and 2glycerolipids）may positively mediate the association between f T3/f T4 ratio and GDM risk,and the mediated proportions were ranged from 20.6%to 47.6%.Conclusions:The association between f T4 levels in early pregnancy and GDM risk was reverse J-shaped,while f T3/f T4 ratio was linearly and positively associated with the risk of GDM.Specific glycerophospholipids and glycerolipids metabolites may partially mediate the positive association between f T3/f T4 ratio and GDM risk,and no significant lipid mediators were observed between f T4 and GDM.