Table_4_Mendelian randomization analysis to investigate the gut microbiome in oral and oropharyngeal cancer.xlsx

BackgroundEvidence supports an observational association between the gut microbiome and susceptibility to extraintestinal cancers, but the causal relationship of this association remains unclear.
MethodsTo identify the specific causal gut microbiota of oral and oropharyngeal cancer, we performed two-sample Mendelian randomization (MR) analysis of gut microbiota on oral and oropharyngeal cancer using a fixed-effects inverse-variance-weighted model. Gut microbiota across five different taxonomical levels from the MiBioGen genome-wide association study (GWAS) were used as exposures. Oral cancer, oropharyngeal cancer and a combination of the two cancers defined from three separate data sources were used as the outcomes. Odds ratios (ORs) and 95% confidence intervals (CIs) for disease per standard deviation (SD) higher abundance of microbiome.
Results & ConclusionsThere was little evidence for a causal effect of gut microbiota on oral and oropharyngeal cancer when using a genome-wide p-value threshold for selecting instruments. Secondary analyses using a more lenient p-value threshold indicated that there were 90 causal relationships between 58 different microbial features but that sensitivity analyses suggested that these were possibly affected by violations of MR assumptions and were not consistent across MR methodologies or data sources and therefore are also to unlikely reflect causation. These findings provide new insights into gut microbiota-mediated oral and oropharyngeal cancers and warrant further investigation.

背景：现有研究证据支持肠道微生物组（gut microbiome）与肠外癌症易感性之间存在观察性关联，但该关联的因果关系仍不明确。方法：为明确口腔癌及口咽癌的特异性致病肠道菌群，本研究采用固定效应逆方差加权模型，以MiBioGen全基因组关联研究（genome-wide association study, GWAS）中5个不同分类层级的肠道微生物组作为暴露因素，开展肠道微生物组对口腔癌及口咽癌的两样本孟德尔随机化（Mendelian randomization, MR）分析。本研究将三个独立数据源定义的口腔癌、口咽癌以及二者的合并队列作为结局指标，计算每1个标准差（standard deviation, SD）更高丰度的微生物组对应的疾病比值比（odds ratios, OR）及95%置信区间（confidence intervals, CIs）。结果与结论：当采用全基因组P值阈值筛选工具变量时，几乎无证据表明肠道微生物组对口腔癌及口咽癌存在因果效应。采用更宽松的P值阈值进行的二次分析显示，58种不同微生物特征间存在90个因果关联，但敏感性分析提示这些结果可能因违反孟德尔随机化假设而存在偏倚，且在不同孟德尔随机化方法及数据源间一致性较差，因此不太可能反映真实的因果关系。本研究为理解肠道微生物组介导的口腔癌及口咽癌发生机制提供了新视角，有待开展进一步探究以验证相关发现。