| Objective:(1)Part Ⅰ: To analyze the association between traditional factors(demographic and clinical characteristics)and gestational diabetes mellitus(GDM)based on the established prospective cohort study,and to screen the high risk factors for establishing the risk prediction model of GDM.(2)Part Ⅱ: Genome-wide association study(GWAS)on gestational diabetes mellitus,obtain and annotate genome-wide single nucleotide polymorphism(SNP)loci,and select target SNP loci that can be used to predict gestational diabetes mellitus.(3)Part Ⅲ:Based on the data from prospective cohort studies,traditional factors(demographic and clinical characteristics)and genetic factors related factors(SNPS found in GWAS)were comprehensively considered to establish a GDM risk prediction model suitable for pregnant women,and the model was evaluated.Methods:(1)Part Ⅰ: A total of 1552 pregnant women with regular antenatal check-up were included in the study from November 2021 to October 2022,The outpatient department of The First People’s Hospital of Shuangliu District of Chengdu was used as the survey field,The demographic and clinical characteristics of the two groups were compared,and the correlation analysis of traditional risk factors of gestational diabetes was further conducted by using generalized propensity score weighting method to screen the risk factors of gestational diabetes.(2)Part Ⅱ: GWAS of gestational diabetes mellitus:two-stage study design.In the first stage,132 patients with gestational diabetes mellitus and 132 normal pregnant women were randomly selected for genome-wide SNP loci chip detection.SNP markers obtained by the whole genome SNP chip were combined for association analysis to explore SNP loci closely related to GDM.In the second stage,the target SNP loci selected by genome-wide association study will be verified in the established cohort by Mass ARRAY,and the SNP loci associated with gestational diabetes will be finally identified,and then the mutation loci will be located to genes to screen candidate pathogenic genes of gestational diabetes.(3)Part Ⅲ:On the basis of previous studies,the SNP loci found in GWAS were combined with traditional risk factors,and variable was selected based on Lasso,while Logistic regression and machine learning methods(support vector:machine-linear kernel and polynomial kernel,Random Forest)were adopted to construct and verify the GDM risk prediction model,select the optimal model to evaluate its ability to predict gestational diabetes mellitus and the feasibility of clinical application.Results:(1)Part Ⅰ: The proportion of diabetic family history,abortion and cesarean section were higher in gestational diabetes patients.The average age and body mass index of pre-pregnancy of gestational diabetes patients were higher than normal pregnant women,and the weight gain in the third trimester and total weight gain During pregnancy of gestational diabetes patients were lower than that in normal pregnant women.The diastolic blood pressure,white blood cell,neutrophils,fasting blood glucose in the first trimester,cystatin C,total cholesterol and triglyceride of gestational diabetes mellitus patients in early pregnancy were significantly higher than those of normal pregnant women.In terms of pregnancy outcomes,gestational week of delivery in gestational diabetes patients was shorter than that of normal pregnant women,the proportion of threatened uterine rupture was higher than that of normal pregnant women,and the proportion of fetal distress was lower than that of normal pregnant women.Analysis of the correlation between the above clinical characteristics of gestational diabetes was carried out,and it was found that age,family history of diabetes and Fasting blood glucose at 8-12 weeks were closely related to the occurrence of gestational diabetes.The risk of GDM in women over 35 years old was 3.569 times that of women under 25 years old.Women with a family history of diabetes were 3.875 times more likely to have gestational diabetes than women without a family history of diabetes.The risk of GDM increased 2.18 times for every 1mmol/1 increase in fasting blood glucose during 8-12 weeks of pregnancy.In addition,The correlation analysis also found that the history of cesarean section,leukocyte and total cholesterol levels in the first trimester were correlated with the occurrence of GDM.(2)Part Ⅱ: Based on the two-stage study design of GWAS of the cohort study.The rs10830963 locus of MTNR1 B gene and rs4939551 locus of CTIF gene were associated with the incidence of GDM in Chinese pregnant women.The risk of GDM was different among different genotypes of rs10830963.The risk of GDM in women with GC genotype was 3.221 times that of women with CC genotype,and the risk of GDM in women with GG genotype was 2.556 times that of women with CC genotype.(3)Part Ⅲ:Gestational diabetes risk prediction model based on traditional factors and genetic factors(SNP found by GWAS)was constructed,A total of three GDM risk prediction models were constructed.After comparison,the optimal model was selected.The model presented as follows: age + body mass index of the pre-pregnancy + family history of diabetes +white blood cell + total cholesterol + diastolic blood pressure + fasting blood glucose(8-12 weeks of pregnancy)+rs10830963;The model obtained a high area under the receiver operating characteristic curve(AUC)0.816,a sensitivity of 80.1%,and a specificity of 73.8%.Support vector machine-linear method is the best method to build GDM risk prediction model.model including genetic susceptibility SNP can improve the model performance.Conclusion: In this study,we built and validated the early pregnancy prediction model of GDM based on the SNP locis of the genetic susceptibility genes and clinical features of pregnant women.This model can help to screen out people with genetic background at high risk of GDM in the early pregnancy,and facilitate the timely intervention of reasonable diet,exercise and other lifestyle in the early pregnancy to reduce the risk of GDM. |
PDF Full Text Request