The vaginal microbiome of reproductive-age women is an indicator of cervical human papillomavirus infection

Background: The understanding of the role played by vaginal microbiome on reproductive health has gained prominence in literature recently, specially regarding to its association to human papillomavirus (HPV) infection. Several risk factors for cervical HPV infection were already identified, highlighting bacterial vaginosis (BV). Despite the association between HPV and BV is well recognized, most aspects of this relation remain unknown. This is in part due to the great heterogeneity present in vaginal environment. Although the recent vaginal microbiome characterization has allowed the understanding of the real microbial composition in that local, the relation between vaginal microbiome components and HPV infection is to be elucidated. Aim: The aim of this study it to evaluate the association between vaginal microbiome components and cervical high-risk HPV (hrHPV) infection. Patients and Methods: We cross-sectionally enrolled 546 women from the five macro-regions of Brazil in previous study entitled “Characterization of the vaginal microbiome in Brazilian women of reproductive-age”. Vaginal microbiome was then determined by sequencing V3-V4 regions of bacterial rRNA 16S gene through MiSeq Platform 300PE (Illumina, San Diego, CA) and clustered in community-state types (CSTs). Cervical samples were subjected to HPV detection and genotyping using Linear Array HPV genotyping kit (Roche Molecular Systems, Pleasanton, CA). We used stepwise (forward, p<0.15) logistic regression to construct two linear scores to predict hrHPV positivity: one based exclusively on the presence of individual bacterial taxa (microbiom-base [MB] score) and the other exclusively on participants’ sociodemographic, behavioral and clinical (SBC) characteristics. The MB score combined coefficients of 30 (out of 116) species retained in the model. The SBC score retained six (age, marital status, new sex partner, hormonal contraceptive use, body mass index and smoking) out of 25 candidate variables. We constructed receiver operating characteristic curves for the scores as hrHPV correlates and compared the areas under the curve (AUC) and 95% confidence intervals (CI) to infer the difference in predictive performance. Results: The prevalence of hrHPV was 15.8% (n=86), and 143 (26.2%) participants had Lactobacillus-depleted vaginal microbiome. The AUCs were 0.8022 (CI: 0.7517-0.8527) for the MB score and 0.7027 (CI: 0.6419-0.7636) for the SBC score (P=0.0163 for the difference between AUCs). Conclusions: Our findings demonstrate the strong correlation between vaginal microbiome components and 10 hrHPV positivity, here demonstrated by MB score, warranting further validation of its clinical utility as a hrHPV infection predictor via longitudinal studies. (AU)