Back

Application of Mendelian randomization to appraise causality in relationships between the gut microbiome and cancer

Dawes, A. C.; Hatcher, C. A.; Mckinlay, A.; Wade, K. H.

2026-01-05 epidemiology
10.64898/2026.01.05.26343428 medRxiv
Show abstract

Cancer is among the leading causes of death worldwide; however, with incidence rising, there is a requirement to identify novel risk factors and biomarkers to prevent and diagnose cancer earlier. Previous work has implicated a role of the gut microbiome in cancer aetiology; however, its causal relevance is not clear. Whilst Mendelian randomisation (MR) is increasingly being applied to assess the causal role of the gut microbiome on several health outcomes, the complexities and limitations of the method are not fully addressed in much of the current literature. Here, we aimed to appropriately apply MR to understand the role of the gut microbiome in cancer aetiology, whilst demonstrating the range of sensitivity analyses required to provide appropriate causal inference. Whilst findings initially suggested that there were several bacterial genera, families and phyla that may alter cancer risk, further sensitivity analyses indicated that these results were unlikely to reflect causality given violations of core MR assumptions. This study highlights the need for a rigorous MR analysis pipeline, incorporating sensitivity analyses evaluating the plausibility of MR-derived effect estimates, to avoid incorrect conclusions around causality, where estimates likely reflect more complex relationships.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.