Species‐level characterization of the cervicovaginal microbiota and its role in human papillomavirus‐associated cervical carcinogenesis

Abstract

AbstractThe cervicovaginal microbiome may contribute to human papillomavirus (HPV)‐associated cervical carcinogenesis, but studies have been limited by low‐resolution analysis methods. Using a high‐resolution bioinformatics pipeline, we evaluated the relationship of the cervicovaginal microbiome with HPV and cervical intraepithelial neoplasia (CIN). The cervicovaginal microbiome of 186 women was characterized by sequencing 16S rRNA regions (V3–V4 and V5–V6) and annotated with the high‐resolution ANCHOR pipeline. Samples were genotyped for HPV using the Roche‐Cobas 4800 assay. We fitted logistic regression models using stepwise forward selection to select species (presence/absence) as correlates of CIN1+ and constructed a linear microbiome‐based score using the regression coefficients. An HPV‐based score was calculated from a separate logistic regression model to detect CIN1+ . Receiver operating characteristic curve analyses were performed; the area under the curve (AUC) and 95% confidence intervals (CI) were compared between scores. Overall, 66.7% of participants were HPV‐positive. 77 unique species were identified: 8 using V3–V4, 48 using V5–V6, and 21 shared. Twelve species were retained via stepwise selection. The AUCs for the microbiome‐, and HPV‐based scores were 0.7656 (95% CI 0.6885–0.8426), and 0.7529 (95% CI 0.6855–0.8204), respectively. Bacterial species may be involved in cervical carcinogenesis as the microbiome‐ and HPV‐based scores performed similarly for CIN1+ detection.