Department of Gynecology and Obstetrics, Tianjin Medical University General Hospital, Tianjin, China.
Tianjin Key Laboratory of Female Reproductive Health and Eugenic, Department of Gynecology and Obstetrics, Tianjin Medical University General Hospital, Tianjin, China.
Front Cell Infect Microbiol. 2023 Jan 19;13:1024723. doi: 10.3389/fcimb.2023.1024723. eCollection 2023.
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 recently published clustering approach called VAginaL community state typE Nearest CentroId clAssifier to assign the metadata to 13 community state types (CSTs) in our study. Nine subCSTs were identified. A random forest model (RFM) classifier was constructed to identify 33 optimal genus-based and 94 species-based signatures. Confounder analysis revealed confounding effects on both study- and hypervariable region-associated aspects. After adjusting for confounders, multivariate analysis identified 14 significantly changed taxa in CC versus the controls ( < 0.05). Furthermore, predicted functional analysis revealed significantly upregulated pathways relevant to the altered vaginal microbiota in CC. Cofactor, carrier, and vitamin biosynthesis were significantly enriched 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.
微生物群相关特征已被用于研究人乳头瘤病毒相关宫颈癌（CC），但由于研究和方法学的异质性，缺乏一致性。本研究分析了四个公开的 16S rRNA 数据集，包括 171 个阴道样本（51 例 CC 与 120 例健康对照），以确定可重现的 CC 相关微生物特征。我们采用了一种新发表的聚类方法，称为 VAginaL 社区状态 typE 最近中心分类器，将元数据分配到我们研究中的 13 种社区状态类型（CST）中。确定了 9 种亚 CST。构建随机森林模型（RFM）分类器，以确定基于 33 个最佳属和 94 个基于种的特征。混杂因素分析显示，混杂因素对研究和高变区相关方面都有影响。在调整混杂因素后，多变量分析确定了 CC 与对照组相比有 14 个显著改变的分类群（<0.05）。此外，预测功能分析显示，与 CC 阴道微生物群改变相关的途径显著上调。CC 中显著富集的途径包括共因子、载体和维生素生物合成，其次是脂肪酸和脂质生物合成以及短链脂肪酸的发酵。差异功能丰度的属贡献也显示出来。总的来说，这项综合研究确定了 CC 中可重复和可推广的特征，表明特定分类群在 CC 发病机制中起因果作用。
