Damon Runyon Cancer Research Foundation reposted this
Does cancer have a microbiome? The field has swung between "yes, every cancer type" and "no, it's all contamination" for years. Last week in Cell by Cell Press, we published an analysis that helps settle it: https://lnkd.in/ef47tuPN The short answer: it depends on where the tumor lives. The result is surprisingly intuitive. Cancers arising at body sites that normally harbor microbiota (oropharyngeal, esophageal, gastric, colorectal) have abundant microbiomes, but in most other cancer types, the microbes we detect are not distinguishable from background contamination. Getting to this answer required a new approach. That's because the tumor microbiome is exceedingly difficult to study, due to low biomass and pervasive contamination. We did three things differently: 1. We developed PathSeq-T2T, which takes a "scorched earth" policy to filtering out human sequences that produce false positives. 2. We assume any given species is contamination unless we have compelling evidence to the contrary. 3. We benchmark microbial detection using in-silico and in-vitro "spike-in" experiments, containing human-microbe mixtures at predefined concentrations. Within orodigestive cancers, microbial communities were highly diverse – often containing multiple species including bacteria, fungi, viruses, archaea, and in a few cases, the protozoan parasite Trichomonas. These communities varied significantly by cancer type and by tumor subtype. Among the most striking findings was a link between tumor mutation burden and microbial colonization, especially by oral-typical microbes. This suggests microbiota preferentially colonize tumors with greater neoantigen load, and that colonization is likely influenced by the immune landscape of the tumor. We also saw a 10-fold depletion of Akkermansia muciniphila in early-onset colorectal cancers, although this result will need to be verified in larger cohorts. In summary, there *is* a cancer microbiome, but it is mostly limited to certain cancer types. In some cases this reflects normal biogeography; in others, tumor- and subtype-specific colonization. There are a few important caveats to add here. Our limit of detection is about 1 microbe per 7k tumor cells, so we won't be able to identify microbes below that level. Our results rely on bulk WGS data, and require validation with orthogonal detection methods. Also, our cohort didn't include cervical cancers (which typically have HPV). Microbial signal was also ambiguous in skin cancer, as many contaminants are skin-associated. We hope this helps guide future cancer microbiome analyses. I also want to acknowledge work by Abraham Gihawi, Yuchen Ge, and Bassel Ghaddar whose publications over the last year set the stage for this paper, as well as Cancer Grand Challenges and the Damon Runyon Cancer Research Foundation for supporting this work. Curious to know what the field thinks, especially how we can further develop standards for tumor microbiome detection.