AUTHOR=Wu Ming , Yu Hongfei , Gao Yueqian , Li Huanrong , Wang Chen , Li Huiyang , Ma Xiaotong , Dong Mengting , Li Bijun , Bai Junyi , Dong Yalan , Fan Xiangqin , Zhang Jintian , Yan Ye , Qi Wenhui , Han Cha , Fan Aiping , Xue Fengxia 

TITLE=Leveraging 16S rRNA data to uncover vaginal microbial signatures in women with cervical cancer

JOURNAL=Frontiers in Cellular and Infection Microbiology

VOLUME=Volume 13 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/cellular-and-infection-microbiology/articles/10.3389/fcimb.2023.1024723

DOI=10.3389/fcimb.2023.1024723

ISSN=2235-2988

ABSTRACT=Microbiota-relevant signatures have been investigated for human papillomavirus-related cervical cancer (CC), but lack consistency because of study- and methodology-derived heterogeneities. Here, four publicly available 16S rRNA datasets including 171 vaginal samples (51 CC versus 120 healthy controls) were analyzed to characterize reproducible CC-associated microbial signatures. We employed a novel clustering approach based on the VAginaL community state typE Nearest CentroId clAssifier to assign the metadata to 13 community state types (CSTs). Twelve subCSTs were identified. A random forest model (RFM) classifier was constructed to identify nine optimal genus-based signatures (Clostridium, Prevotella, Wolbachia, Streptococcus, Rhodococcus, Campylobacter, Barnesiella, Peptostreptococcus, and Moryella), with high accuracy for discriminating CC from the controls (area under the receiver operating characteristic curve — 82.06% on the training set, 81.08% on the test set). Confounder analysis revealed confounding effects on both study- and hypervariable region-associated aspects. After adjusting for confounders, multivariate analysis identified five significantly enriched genera in CC versus the controls (P < 0.05), including Barnesiella, Peptostreptococcus, Clostridium, Bacteroides, and Streptobacillus; three of these were signatures revealed by the RFM (Barnesiella, Peptostreptococcus, and Clostridium). Furthermore, predicted functional analysis revealed significantly upregulated pathways relevant to the altered vaginal microbiota in CC. Cofactor, carrier, and vitamin biosynthesis were significantly elevated in CC, followed by fatty acid and lipid biosynthesis, and fermentation of short-chain fatty acids. Genus-based contributors to the differential functional abundances were also displayed. Overall, this integrative study identified reproducible and generalizable signatures in CC, suggesting the causal role of specific taxa in CC pathogenesis.