Endocrine Abstracts

Background: The gestational diabetes mellitus (GDM) is a metabolic disorder characterized by glucose intolerance in pregnant women without pre-existing diabetes, typically diagnosed in mid to late pregnancy. The global prevalence of GDM is on the rise, posing significant diagnostic and economic burdens on society and affecting the health of two generations, potentially impacting population quality. Early diagnosis or understanding of the pathophysiology of this disease in early pregnancy may effectively reduce its incidence. Emerging evidence indicates that the gut microbiome can modulate metabolic homeostasis, thereby influencing the development of GDM. However, it remains uncertain whether and how the gut microbiota and its blood-related metabolites change in GDM.

Methods: We collected serum samples from 698 pregnant women, including 190 in mid-pregnancy and 508 in early pregnancy, as well as partial fecal samples. Utilizing a combination of untargeted serum metabolomic analysis using liquid chromatography-mass spectrometry and fecal metagenomic sequencing, we identified significant changes in gut microbiome-associated metabolite abundance in GDM patients and matched controls. Subsequently, we developed diagnostic and early prediction models for GDM based on different modeling methods, which were evaluated in independent validation and other study cohorts.

Results: Our study revealed significant changes in 1024 serum metabolites in GDM patients, with 275 metabolites closely related to the gut microbiota. Among them, five metabolites showed significant differences in the early pregnancy stages of GDM development. Through repeated testing using untargeted metabolomics, these five metabolites could effectively distinguish GDM patients from normal individuals and even identify patients at risk of developing GDM. Constructing different diagnostic and predictive models based on these five metabolites yielded the highest AUC of 0.95 in the testing cohort and AUCs of 0.91 (sensitivity of 81.5%, specificity of 85.2%) and 0.88 (sensitivity of 79.5%, specificity of 82.4%) in the validation cohorts. Furthermore, the model demonstrated significant advantages in sensitivity, specificity, and accuracy in other studies.

Conclusion: The microbiota and serum metabolites of GDM patients significantly differ from those of matched controls, and the reprogramming of the gut microbiome in GDM patients is related to changes in the serum metabolome. The use of effective serum metabolite models can advance the prediction of GDM to early stages, thereby reducing the adverse consequences of GDM.