A prospective cohort study examining the association between the periconceptual vaginal microbiota and first-trimester miscarriage in Kenyan women

Background: Studies evaluating the association between the vaginal microbiota and miscarriage have produced variable results. Objective: This study evaluated the association between periconceptual and first-trimester vaginal microbiota and women's risk for miscarriage. Methods: At monthly preconception visits and at 9-12 weeks gestation, women collected vaginal swabs for molecular characterisation of the vaginal microbiota. Participants who became pregnant were followed to identify miscarriage versus pregnancy continuing to at least 20 weeks gestation. Results: Forty-five women experienced miscarriage and 144 had pregnancies continuing to >= 20 weeks. A principal component analysis of periconceptual and first-trimester vaginal bacteria identified by 16S rRNA gene PCR with next-generation sequencing did not identify distinct bacterial communities with miscarriage versus continuing pregnancy. Using taxon-directed quantitative PCR assays, increasing concentrations of Megasphaera hutchinsoni, Mageeibacillus indolicus, Mobiluncus mulieris and Sneathia sanguinegens/vaginalis were not associated with miscarriage. In exploratory analyses, these data were examined as a binary exposure to allow for multivariable modelling. Detection of Mobiluncus mulieris in first-trimester samples was associated with miscarriage (adjusted relative risk [aRR] 2.14, 95% confidence interval [CI] 1.08, 4.22). Additional analyses compared women with early first-trimester miscarriage (range 4.7-7.3 weeks) to women with continuing pregnancies. Mobiluncus mulieris was detected in all eight (100%) first-trimester samples from women with early first-trimester miscarriage compared to 101/192 (52.6%) samples from women with continuing pregnancy (model did not converge). Detection of Mageeibacillus indolicus in first-trimester samples was also associated with early first-trimester miscarriage (aRR 4.10, 95% CI 1.17, 14.31). Conclusions: The primary analyses in this study demonstrated no association between periconceptual or first-trimester vaginal microbiota and miscarriage. Exploratory analyses showing strong associations between first-trimester detection of Mobiluncus mulieris and Mageeibacillus indolicus and early first-trimester miscarriage suggest the need for future studies to determine if these findings are reproducible.