Modeling the Vaginal Microbiome: Predicting BV Recurrence and Identifying Preventative Targets

Bacterial vaginosis (BV) is the most common form of vaginitis, affecting approximately one in three women at some point in their lives. Women with BV face serious and costly reproductive and obstetric complications, including the acquisition of sexually transmitted infections such as HIV, preterm delivery, miscarriage, infertility, and pelvic inflammatory disease. Characterized by a lack of Lactobacillus spp. in the vagina and the presence of a diverse, anaerobic microbiota, BV is diagnosed by three of four Amsel’s criteria: vaginal pH > 4.5, presence of clue cells, fishy odor with 10% KOH, and abnormal discharge. Symptoms include malodor, vaginal discharge, irritation, itching, or painful urination, which can severely impact a woman’s self-esteem and personal life. Currently, there is no long-term cure for BV. Antibiotic treatment is recommended when symptoms are reported, primarily to provide relief. However, recurrence rates are high: 60% at six months, and 70% one year later. Notably, the risk of gynecological complications exists regardless of the presence of symptoms. Significant gaps in our understanding of the pathogenesis and etiology of BV have led to its heterogeneous definition, imprecise diagnostic methods, non-curative treatments, and numerous co- morbidities. The vaginal microbiome, measured with whole-genome shotgun metagenomic sequencing, and our novel approach, metagenomic Community State Types (mgCSTs), have identified eight functionally distinct types of BV. Preliminary data from our case-control study on BV recurrence demonstrate that one mgCST is highly associated with BV recurrence six months following an initial BV event, indicating that a strain composition- based approach is transforming our understanding of BV. The central hypothesis is that using mgCSTs or other metagenome-based features, instead of the historically heterogeneous definition of BV, will clarify the molecular mechanisms behind its recurrence and identify new therapeutic targets. We aim to: (1) employ metagenome- assembled genome-wide association analyses and a novel statistical method to uncover mechanisms by which the vaginal microbiome facilitates the spontaneous resolution of BV, and (2) develop a predictive model of BV recurrence from metagenomic features of the vaginal microbiome, a dataset ideally suited for innovative machine learning methods for development of a predictive model. These aims will reveal innovative and testable hypotheses regarding how the vaginal microbiome facilitates BV resolution, serving as potential targets for developing therapeutic interventions to treat recurrent BV. Furthermore, the hypothesized mechanisms developed in Aim 1 will serve as the foundation for future in vitro studies validating the efficacy of potential therapeutic targets. The potential of Aim 2 is the development of a highly predictive model of BV recurrence that will undergo validation in a subsequent study laying the foundation for an innovative diagnostic assay for BV. The successful completion of this proposal will lay the groundwork toward a revolution in personalized medicine, maximizing women’s gynecological and obstetric health. Project Number: 1R21AI196307-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of Allergy and Infectious Diseases (NIAID) | Principal Investigator: Johanna Holm | Institution: UNIVERSITY OF MARYLAND BALTIMORE, BALTIMORE, MD | Award Amount: $427,625 | Activity Code: R21 | Study Section: Interspecies Microbial Interactions and Infections Study Section[IMII] View on NIH RePORTER: https://reporter.nih.gov/project-details/1R21AI19630701