Next Generation Metabolic Signatures of Infectious Causes of Vaginitis: Small Molecule Metabolite-Based Biomarkers for Diagnosis, Test-of-Cure, and Prediction of Recurrence

About This Grant

Abstract PROJECT SUMMARY/ABSTRACT In the US, vaginitis accounts for greater than 10 million medical office visits each year. Vaginal discharge syndrome is often indicative of bacterial vaginosis (BV), vulvovaginal candidiasis (VVC), or trichomoniasis. Each condition is associated with adverse reproductive and sexual health sequelae; annual costs for managing vaginitis exceed several billion dollars. Accurate diagnosis of vaginal discharge is critical for providing appropriate therapy. Some diagnostic tests are rapid and inexpensive, but insensitive, such as saline wet mount with microscopy to diagnose trichomoniasis. Other tests are highly sensitive and specific, such as nucleic acid amplification tests, but are expensive and may delay the diagnosis if sent offsite to a reference lab. A substantial number of women with vaginitis are misdiagnosed and receive inappropriate treatment. Syndromic management is common and there is an unmet need for new approaches of diagnosing infectious causes of vaginitis. Small molecule metabolites in vaginal fluid, products of microbial and host metabolism, have great potential to serve as diagnostic biomarkers for vaginitis. We have shown that metabolic profiles are dramatically different in women with and without BV and have identified several individual metabolites that have diagnostic potential. These molecules may also be useful for early detection, prediction of recurrence, and evaluation of test-of-cure. We propose to identify and validate vaginal fluid metabolites that may be diagnostic of BV, trichomoniasis and WC (Aim 1). We will also evaluate changes in concentrations of metabolites prior to the onset of symptoms to assess early predictors of disease and determine if changes in metabolites after treatment of vaginitis predict subsequent relapse (Aim 2). We will use vaginal fluid samples from two study populations for the training set. The first set includes samples previously collected from 251 women with >100 episodes of vaginitis. A unique feature is that daily samples were collected both prior to diagnosis of vaginitis and post-treatment. The second set will comprise prospectively collected vaginal fluid samples from 105 women with and without vaginal symptoms from a different geographical site to ensure selection of representative biomarkers. A sequential metabolomics approach will be used. First, a global untargeted metabolomics screen will provide >4000 metabolite features. A subset of metabolites will be selected using machine learning approaches that accurately classify each vaginitis condition. Absolute concentrations of this subset will be measured, and diagnostic performance will be assessed when compared with gold standard tests. High-performing metabolites will be used to develop a diagnostic algorithm that will combine metabolites for optimal diagnostic performance. The diagnostic algorithms will be independently validated using vaginal fluid samples from 720 women with and without vaginal discharge attending 26 sites in 18 US Metropolitan Statistical Areas. We will also determine the value of these metabolites for test-of-cure and ear1y diagnosis. We expect that this study could result in a simple, streamlined diagnostic algorithm for vaginitis based on metabolites produced by vaginal pathogens.