Abstract: Objective  To investigate the early prediction efficacy of adiponectin (ADPN) for gestational diabetes mellitus (GDM), and to explore new indicators for the early diagnosis of GDM and risk models for early prediction. Methods  A cohort of 486 pregnant women in early pregnancy (7－12 weeks) was selected from July to November 2023 at Beijing Tongren Hospital, Capital Medical University. According to the diagnostic criteria of GDM recommended by the International Association for the study of Diabetes and Pregnancy Study Group (IADPSG) in 2010,  mid-pregnancy pregnancies were divided into GDM group (150 cases) and non-GDM group (336 case). ADPN, insulin (IR), fasting glucose (GLU), and glycated albumin (GA) were collected in early pregnancy, and the  homeostatic model assessment of adiponectin (HOMA-AD), homeostatic model assessment of insulin resistance index (HOMA-IR) and hepatic steatosis index (HSI) were calculated. The differences in ADPN, HOMA-AD, and HOMA-IR between the two groups were analyzed and compared, and the value of each type of index in predicting GDM was analyzed with the receiver operating characteristics (ROC) curve, and the predictive risk model was established by combining the relevant indexes.  Results  There was a statistically significant difference between the GDM and non-GDM groups in ADPN in early pregnancy (P<0.05). The results of the ROC curve analysis showed that the area under the curve (AUC) of ADPN for early prediction of GDM positivity was 0.723, with a cutoff value 13.38 mg/L. There was a statistically significant difference between the GDM and non-GDM groups in HOMA-AD (P=0.000). The AUC of HOMA-AD for early prediction of GDM was 0.815, with the cutoff value 3.024. Combining GLU, HOMA-AD, HOMA-IR, and HSI in a Logistic regression model improved predictive performance across several metrics, with the final test set of AUC=0.829, accuracy=0.740, sensitivity=0.913, negative predictive value=0.833. Conclusion  ADPN levels were reduced in the GDM group compared to the non-GDM group, and the diagnostic efficacy of a single ADPN was poor when it was used for early prediction of the onset of GDM. The HOMA-AD level of the GDM group was lower than that of the non-GDM group, and HOMA-AD was negatively correlated with the disease, which was more effective than ADPN, HOMA-IR, and HIS in the early prediction of GDM. HOMA-AD could be used in combination with these indexes to establish a diagnostic and predictive model to improve the effectiveness of the prediction.

Key words: adiponectin, HOMA-AD, gestational diabetes mellitus, early prediction, establishment of risk model