The Utility of Voided Urine Samples as a Proxy for the Vaginal Microbiome and for the Prediction of Bacterial Vaginosis

Abstract

Abstract Recent work has shown that the vaginal microbiome exerts a strong impact on women's gynecological health. However, collection of vaginal specimens is invasive and requires previous clinical training or the involvement of a trained clinician. In contrast, urine sample collection is routine and noninvasive and does not require involvement of a clinician. We sought to compare the vaginal and urogenital microbiomes to assess the utility of voided urine samples as a proxy for the vaginal microbiome. Paired urogenital and vaginal samples were collected from pregnant women and characterized by 16S rRNA taxonomic profiling. We examined diversities and compositions of paired urogenital and vaginal microbiomes using five discrete strategies to explore the similarity between the vaginal and urogenital microbiomes. A strategy comparing the paired urogenital and vaginal microbiomes in which taxa were assigned using the STIRRUPS database and urine-specific taxa were removed showed no significant difference in diversity and composition between the paired urogenital and vaginal microbiomes. Moreover, the relative abundances of common vaginal taxa were linearly correlated with those in the paired urogenital microbiomes. These similarities suggest that voided urine samples could represent a noninvasive protocol for accurate profiling of the vaginal microbiome with likely clinical applications. Finally, a machine learning model was established in which the voided urine microbiome was compared favorably to the vaginal microbiome in predicting bacterial vaginosis.