Early-pregnancy trunk phase angle derived from bioelectrical impedance analysis for the prediction of gestational diabetes mellitus.

OBJECTIVE: To evaluate first-trimester predictors of gestational diabetes mellitus (GDM) and develop an early risk model. METHODS: Women at 8-12 weeks of gestation who attended routine antenatal care at Hainan General Hospital between October 2023 and December 2024 were enrolled and followed until 24-28 gestational weeks for a 75-g oral glucose tolerance test (OGTT). Participants were classified into a GDM group or a normal glucose tolerance (NGT) group according to standard diagnostic criteria. General clinical characteristics, fasting biochemical parameters, and body-composition variables were collected in early pregnancy; Body composition was measured at 13 weeks using a multi-frequency bioelectrical impedance analyzer (InBody 770). Logistic regression, ROC, bootstrap validation, calibration plots, and decision curve analysis were used. RESULTS: Of 125 women (68 GDM, 57 NGT), GDM had higher age, pre-pregnancy body mass index (p-BMI), assisted conception, family hyperglycemia, first-trimester weight gain, glucose, insulin, HOMA-IR, TyG, lipids, neutrophils, uric acid, and most body-composition indices (all P<0.05). After adjustment, 25 variables (P<0.01) were grouped into six domains. HOMA-IR, triglycerides (TG), first-trimester gestational weight gain (GWG-F), and 50-kHz trunk phase angle (trunk PhA) were selected as core predictors. The BIC-based model identified HOMA-IR (OR = 3.287), TG (OR = 2.277), GWG-F (OR = 1.150), trunk phase angle (OR = 2.615), and age (OR = 1.180) as independent risk factors. The model showed good discrimination (AUC = 0.921), calibration (Hosmer-Lemeshow P = 0.287), and net benefit. CONCLUSION: HOMA-IR, TG, GWG-F, trunk PhA, and age are independent early GDM predictors. The combined model (AUC = 0.921) can help identify high-risk women early for timely intervention.