Synergistic use of 1,5-AG and HbA1c for early prediction of gestational diabetes: capturing BMI-dependent glycemic phenotypes

Abstract Purpose Recognizing metabolic heterogeneity in gestational diabetes mellitus (GDM) and body mass index (BMI)-linked phenotypes, we evaluated whether combining hemoglobin A1c (HbA1c, reflecting fasting glycaemia) and 1,5-anhydroglucitol (1,5-AG, reflecting post-load glucose excursions) improves early prediction and whether performance differs by BMI. Methods In this multicenter retrospective study, pregnant women who had 1,5-AG and HbA1c measured before 20 weeks of gestation at two tertiary centers in Japan were included. Spearman’s correlation was used to assess associations between glycemic markers and glucose levels. Predictive performance for GDM was evaluated using ROC analysis, and stratified analyses were conducted by pre-pregnancy BMI. Results Among 191 participants, 45 (24.1%) developed GDM: 35.1 ± 4.9 years, pre-pregnancy BMI 22.9 ± 4.3 kg/m 2 , and sampling at 14.3 [IQR 14.0–14.7] weeks. HbA1c correlated with fasting glucose ( r = 0.35) while 1,5-AG correlated inversely with 2-h glucose ( r = − 0.39). They themselves were not significantly correlated ( r = − 0.13). As single predictors, performance depended on BMI: in ≥ 25.0 kg/m 2 , HbA1c outperformed 1,5-AG (AUC 0.776 vs 0.618); in BMI < 25.0 kg/m 2 , 1,5-AG outperformed HbA1c (AUC 0.704 vs 0.640). In both BMI strata, the dual-marker model was superior (AUC 0.833 and 0.803, respectively) and provided more balanced sensitivity, accuracy, and F1. Pre-pregnancy BMI correlated positively with fasting plasma glucose ( r = 0.47) but not with 1-h or 2-h glucose ( r = 0.20 and r = 0.16, respectively), supporting BMI-related metabolic variation. Conclusion Combining 1,5-AG and HbA1c enhances early prediction of GDM by capturing complementary glycemic abnormalities linked to BMI-specific metabolic phenotypes.