Metagenomic profiling and predictive modeling of the gut microbiome reveal signatures of gestational disease

ABSTRACT The gut microbiome plays a vital role in maternal health and pregnancy outcomes, yet its impact on conditions like gestational hypertension (GH) and gestational diabetes mellitus (GDM) remains poorly understood. This study explores how the gut microbiome differs between pregnant women with these conditions and healthy controls, using metagenomic sequencing to analyze microbial composition and function. Our findings reveal that women with GH and GDM exhibit greater microbiome variability and distinct shifts in bacterial communities compared to healthy pregnancies. Key beneficial bacteria, such as Bacteroides fragilis and Roseburia intestinalis , were reduced in cases, suggesting potential disruptions in gut-related metabolic and immune functions. In addition to multiple differentially abundant species of Sphingobacterium in cases versus controls, functional analysis indicated changes in carbohydrate and lipid metabolism, reinforcing the microbiome’s connection to metabolic health. Furthermore, machine learning models demonstrated promising results in predicting disease status based on microbiome data, underscoring the potential for gut bacteria as potential predictive biomarkers for pregnancy-related conditions. These insights highlight the gut microbiome’s role in pregnancy health and suggest it may be a promising target for future interventions aimed at reducing complications and improving maternal-fetal outcomes. IMPORTANCE Gut microbial dysbiosis has been implicated in pregnancy complications, yet most studies rely on 16S rRNA sequencing, which limits resolution and functional insight. Here, using shotgun metagenomic sequencing and machine learning, we identified robust microbial taxonomic and functional signatures that distinguish gestational hypertension and gestational diabetes from healthy pregnancies. A combined feature set enabled accurate classification of disease status, with overlapping features between statistical and predictive frameworks underscoring biological relevance. Altogether, our study defines high-resolution microbiome signatures with translational potential as predictive biomarkers for maternal health, while also providing an open, reproducible analysis pipeline to support future investigations.