Early pregnancy metabolic disturbances appear to offer meaningful predictive value for GDM in Asian Indian women. Published in Cardiovascular Diabetology, the study evaluated whether untargeted metabolomic signatures could identify GDM risk months before standard screening.

Within the STratification of Risk of Diabetes in Early (STRiDE) pregnancy cohort, plasma samples collected before 16 weeks of gestation were analyzed using untargeted ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS). Samples from 50 individuals who later developed GDM and 50 normoglycemic individuals formed the analytical set. Statistical modelling and machine learning approaches, including logistic regression and random forest (RF), were used to characterize metabolite–disease associations and construct predictive models.

A total of 49 metabolites showed significant associations with GDM, predominantly lipid species such as phosphatidylcholines, sphingomyelins, and triacylglycerols. A focused eight-metabolite panel delivered high predictive accuracy with an AUC of 0.880 (95% CI 0.809–0.951). When combined with conventional clinical risk factors, predictive performance remained comparable (AUC 0.880, 95% CI 0.810–0.952). Pathway enrichment analysis highlighted dysregulation of glycerophospholipid and sphingolipid metabolism, autophagy, and insulin-resistance pathways.

The findings indicate that early-pregnancy metabolomic profiling may support timely stratification of GDM risk in Asian Indian populations.