Multi-omics investigation of metabolomic and microbial features in bacterial vaginosis

Abstract Bacterial vaginosis (BV) affects 23–29% of reproductive-age women globally and recurs in over half of cases within 6 months, yet the coordinated metabolic and genomic mechanisms underlying it remain incompletely understood. In this cross-sectional study of 111 Swedish women (29 with BV, 82 controls), we combined shotgun metagenomics, targeted liquid chromatography–mass spectrometry metabolomics, and Gardnerella metagenome-assembled genome (MAG) analysis, integrating these layers with supervised (DIABLO) and unsupervised (MOFA2) methods. BV status explained 23.5% of community variance and was characterized by Lactobacillus depletion, enrichment of canonical anaerobes including Gardnerella vaginalis and Fannyhessea vaginae , and a pronounced metabolic shift marked by elevated polyamines (cadaverine, putrescine, N-acetylputrescine, trimethylamine) and depleted nucleosides. Supervised and unsupervised integration converged on the same discriminatory features and prioritized Lactobacillus loss and the polyamine producer Dialister micraerophilus over the most abundant BV taxa. Across 35 Gardnerella MAGs spanning nine species, no accessory genes were associated with BV after correction, yet accessory gene content explained 8.5% of metabolome variance independently of BV status ( p = 0.002). These cross-sectional findings are associative and hypothesis-generating, supporting a community-level, metabolically defined model of BV rather than strain-specific pathogenesis.