This study applied a two-sample Mendelian randomisation (MR) framework to evaluate whether the composition of the oral microbiome causally affects risk for cancers of the oral cavity, oropharynx, and tongue, and to distinguish forward from reverse causal effects (i.e., whether cancers alter microbial abundance).
The authors state the primary objective as assessing potential causal relationships between dorsal-tongue and salivary microbial taxa and three cancer outcomes: oral cancer, oropharyngeal cancer, and tongue cancer.
Genetic instruments were derived from genome-wide association study (GWAS) summary statistics obtained from public repositories including CNGBdb and the FinnGen consortium, along with other sources identified by the authors. Analyses were implemented using the R package TwoSampleMR, which enables two-sample MR with summary-level data.
The investigators used single-nucleotide polymorphisms (SNPs) as instrumental variables for microbial taxa. The abstract reports an SNP significance threshold of p < 5 × 10^-6 for instrument selection.
Selected SNPs meeting the reported significance threshold were used as genetic proxies for dorsal-tongue and salivary microbial taxa. The principal MR estimator applied was the inverse-variance-weighted (IVW) method to combine SNP-specific causal estimates into an overall effect size for each microbial taxon on each cancer outcome.
To evaluate the credibility of causal inferences, the authors performed standard MR sensitivity checks including heterogeneity testing and pleiotropy assessments. These checks are intended to detect violations of MR assumptions such as horizontal pleiotropy or heterogeneity among instrument estimates.
Using the described two-sample MR approach and the SNP selection threshold of p < 5 × 10^-6, the study identified genetically supported causal relationships between specific oral microbial taxa (from saliva and tongue) and risk of oral, oropharyngeal, and tongue cancers.
When integrating results across analyses, two taxa were highlighted as associated with a reduced risk of oropharyngeal and tongue cancers: s Veillonella_rogosae_mgs_2008 and s unclassified_mgs_1048. The abstract reports these taxa as conferring reduced cancer risk; however, the abstract does not provide numerical effect estimates, confidence intervals, allele information, or the number of instruments per taxon. Those details would be available in the full article.
The authors report that sensitivity analyses, heterogeneity testing, and pleiotropy evaluations corroborated the robustness of their main MR findings. These analyses were performed to assess whether the IVW estimates could be biased by heterogeneity among SNP effects or by directional pleiotropy.
The abstract indicates that the results remained credible after these assessments, but does not provide test statistics, p values, or the specific sensitivity methods used beyond general descriptors. For a complete appraisal, the full methods and results sections in the manuscript would need to be consulted.
To explore reverse causation, the study also conducted bidirectional MR sensitivity analyses testing whether oral, oropharyngeal, and tongue cancers influence the abundance of specific oral microbes.
Reverse MR analyses indicated that oral and tongue cancers may themselves alter the abundance of particular oral microbial taxa, suggesting a potential bidirectional causal loop between certain microbes and cancer phenotypes. The abstract does not list which taxa were affected in the reverse direction nor the magnitude of those effects.
Based on large-scale publicly available genetic data and two-sample MR analyses, the authors conclude that there are significant causal relationships between the oral microbiota and cancers of the oral cavity, oropharynx, and tongue. Two taxa—Veillonella_rogosae_mgs_2008 and an unclassified taxon labeled mgs_1048—were specifically associated with reduced risk of oropharyngeal and tongue cancers.
The authors recommend future integration of metagenomic data to validate and extend these microbiota–cancer associations. The abstract emphasizes that reverse MR findings imply the possibility of bidirectional interactions in which cancer status may affect microbial abundance.
The abstract documents the study design, data sources, analytical approach, key taxa identified, and general robustness checks, but it does not report numerical effect sizes, confidence intervals, sample sizes, population ancestries, or the full list of taxa evaluated. These items were not reported in the abstract and would require review of the full article for complete details.
Clinicians and researchers interpreting these findings should consult the full text for instrument strength, detailed sensitivity results, and subgroup analyses before translating these genetic associations into mechanistic hypotheses or clinical recommendations.