1234-P: Early Prediction of Gestational Diabetes Based on Novel Biomarkers and Risk Factors in Early and Midpregnancy—A Longitudinal Study

Objectives: We aimed to improve early GDM prediction using a comprehensive biomarker panel in early and mid-pregnancy, considering clinical accessibility. Methods: In a US study of 2802 multi-racial/ethnic pregnant individuals, we assessed 91 cardiometabolic biomarkers at gestational weeks 10-14 (random blood) and 15-26 (fasting) among 107 GDM cases and 214 controls. Biomarkers significantly associated with GDM were selected for prediction and categorized for clinical accessibility: Group I (routine clinical tests like HbA1c, hs-CRP, lipids), Group II (less routinely assessed biomarkers like IGF axis markers, adipokines), Group III (specialty lab-required targeted metabolomics like amino acids (AAs) and phospholipid fatty acids (FAs)). We developed a comprehensive full model incorporating these biomarkers alongside conventional predictors (age, BMI, plasma glucose) and alternative models based on clinical accessibility. Model performance was evaluated by AUC, proportion of cases followed (PCF, %), and proportion needed to follow (PNF, %). Results: At 10-14 GW, the full model had the highest discriminative accuracy, AUC of 0.842. Similarly, the full model at 15-26 GW had the highest accuracy, AUC of 0.829. The addition of novel biomarkers improved clinical applications as demonstrated by PCFs and PNFs. For instance, at 10-14 GW, following women with a GDM risk above 80% as estimated by our models would identify 69.5% of GDM cases in the full model vs. 49.1% in the conventional model. To identify 90% of GDM cases, the proportion of women needed to follow based on the estimated risks would be 46.1% in the full model vs. 71.1% in the conventional model. Conclusion: A selected panel of biomarkers using early-pregnancy random plasma samples can be as predictive as those using mid-pregnancy fasting samples. HbA1c, key IGF axis markers and adipokines, and selected phospholipid FAs and AAs effectively predict GDM risk, offering meaningful clinical utility. Disclosure J. Yang: None. D.B. Sacks: Other Relationship; Sebia, Trinity. M.Y. Tsai: None. J. Chen: None. C. Zhang: None. Funding Dean's Office, Yong Loo Lin School of Medicine, National University of Singapore; The Eunice Kennedy Shriver National Institute of Child Health and Human Development intramural funding and American Recovery and Reinvestment Act funding: HHSN275200800013C, HHSN275200800002I, HHSN27500006, HHSN275200800003IC, HHSN275200800014C, HHSN275200800012C, HHSN275200800028C, HHSN275201000009C, HHSN275201000001Z