Abstract
AbstractBackgroundThe vaginal microbiome is an underexplored environment which may yield insights about endometrial cancer (EC) pathophysiology and serve as an early marker of disease. Our objective was to evaluate the preoperative vaginal microbiome in women undergoing hysterectomy for benign conditions and EC to identify patterns which could segregate benign disease from EC, and to assess if the microbiome distinguishes low-grade (LG) from high-grade (HG) histologies.MethodsVaginal microbiome samples were prospectively collected at the time of hysterectomy, and clinicopathologic data collected. Extracted DNA underwent shotgun metagenomics sequencing. Microbial diversity was calculated using the Shannon Index (α-diversity) and PERMANOVA (β-diversity). Hierarchical clustering was used to describe community state types (CST), which were then compared by microbial diversity and grade. Machine learning was utilized to assess the predictive value of grade and histology based on relative bacterial abundance.Results61 patients participated: 1) benign gynecologic disease (n=11), 2) LG EC (n=30), and 3) HG EC (n=20). 40 (62.5%) were White, 22 (34.3%) were Black, and 37 (57.8%) were Hispanic. Both α- and β-diversity were associated with tumor grade (p□=□0.026 and p□=□0.035, respectively). Four CST were identified that associated with grade of disease (p=0.036). Different histologies demonstrated variation in CST even within tumor grades (p=0.017). Markers at the species level informed models that predicted benign vs cancer (AUC 0.878), HG versus benign (AUC=0.803), and HG versus LG (AUC=0.771) with high accuracy.ConclusionsThe vaginal microbiome segregates benign disease from EC, and is predictive of histology and grade, suggesting it may be an effective tool for screening.