Digestive-cancer microbiomes confirmed in large genome study; most tumors show no distinguishable microbiome

Overview

The Conversation reports on a landmark analysis of cancer genomes that addresses whether tumours carry their own microbiomes. Using Genomics England’s 100,000 Genomes Project, the team built a rigorous pipeline to separate genuine microbial signals from contamination, applying it to more than 16,000 tumours. The findings show that most cancers, including brain, breast and kidney cancers, lack a microbiome distinguishable from background, suggesting previous signals may reflect contamination. However, cancers of the mouth, oesophagus, stomach and bowel display clear microbial communities, including bacteria, viruses, fungi and archaea, with variations by location and cancer subtype.

Key insights and data resources are published to help standardise methods and reduce wasted samples in this field.

• Digestive tract cancers harbor consistent microbial life, while many other cancers do not.
• Contamination can masquerade as microbial signals; rigorous filtering is essential.
• Open data and software enable other researchers to apply the same robust approach.
• The findings could influence cancer screening and treatment strategies where microbes are implicated.

Introduction: addressing the microbiome in cancer

For decades cancer was considered a human disease driven solely by rogue cancer cells. A growing body of research has proposed that tumours could host communities of microbes, potentially affecting cancer growth, treatment resistance, or metastasis. Yet the field suffered from contradictory claims, divergent methods, and at least one notable retraction when findings could not be replicated. The Conversation and colleagues set out to settle the question with a uniquely robust approach, leveraging the world’s largest cancer genetic dataset from Genomics England’s 100,000 Genomes Project, which includes DNA from more than 16,000 tumours. The objective was to determine which cancers truly harbor microbiomes distinguishable from background signals and which do not, while also constructing a pipeline that minimizes false positives from contamination.

Data source and methodology: a rigorous pipeline

The study exploits whole-genome sequencing data from a massive, well-curated cancer cohort. Rather than ignoring the non-human DNA that accompanies human genomes, the researchers systematically filtered out human sequences and compared the non-human reads against curated microbial genome databases to identify microbial signals. This approach is challenging because there is no single definitive human genome; individual variation and gaps in reference assemblies can lead to leftover human sequences being misclassified as microbial hits. Similarly, microbial reference libraries themselves can contain misannotations or laboratory contaminant DNA, which must be accounted for. To address these issues, the team implemented aggressive filtering against multiple versions of the human genome, used up-to-date alignment tools, and cross-checked findings across cancer types to flag species likely introduced during sample handling.

Findings: which cancers carry a microbiome?

The latest analyses reveal a nuanced picture. Most cancers studied—such as brain, breast and kidney cancers—lacked a microbiome that could be distinguished from background contamination. This implies that many previously reported microbial signals in these tumours may reflect contamination from laboratory reagents, equipment, or personnel rather than true intratumoural microbiota. In contrast, tumours arising in the digestive tract—mouth, oesophagus, stomach and bowel—show clear, consistent evidence of microbial life. In these cancers, microbes including bacteria, viruses, fungi, and archaea were detected, with the specific mixture of species varying by location within the digestive tract and by cancer subtype and mutational burden. The researchers even identified protozoan parasites in some samples, illustrating the diversity of microorganisms that can co-exist with cancer cells in these sites.

Contamination: the hardest part and how it was tackled

Disentangling real microbes from contamination proved to be the central challenge. Sequencing a tumour reads all DNA in the sample, human and non-human. The team chose to analyze the non-human reads directly, but this decision amplified the risk that leftover human sequences could masquerade as microbial DNA if they resembled microbial genomes. Problems also arise from errors in microbial reference databases and from skin or laboratory DNA contaminating samples. The researchers tackled these issues with a multi-pronged strategy: filtering against several human genome references, employing curated microbial databases, and using cross-sample comparisons to identify species that appear across many cancer types (likely lab contaminants) versus those restricted to a few cancer types (more likely genuine). They demonstrated that common skin bacteria and other contaminants often appear across all cancers, reinforcing the need for large-scale, high-resolution data to separate signal from noise. This rigorous approach, made feasible by the Genomics England dataset, would be difficult in smaller studies and may explain why earlier results were inconsistent or non-reproducible.

Open science: data and tools for the field

In a move that mirrors best practices in modern computational biology, the authors have released their data and software openly. They provide a downloadable software package and a vetted list of microbial species they are confident are genuinely present in tumours, enabling other researchers to apply the same rigorous contamination-control framework to their own data. The goal is to reset expectations in the microbiome-cancer field, directing research focus toward robust microbial signals in digestive tract cancers and their potential roles in tumour development and treatment response. By standardising methods and providing accessible resources, the study aims to prevent wasted samples and false positives that have hampered progress in the past.

Implications and future directions

These findings carry important implications for cancer screening, diagnostics, and therapy development. If microbial communities in digestive cancers influence tumour behaviour or treatment outcomes, they could become targets for screening or precision therapies. Conversely, the lack of a distinguishable microbiome in most cancers suggests that microbiome-based biomarkers or interventions may be limited to specific cancer types or contexts. The study also underscores the importance of rigorous methodological benchmarks and the need for large, high-quality datasets to separate real biological signals from artefacts of sampling and laboratory processing. The open data and open software pave the way for replication, validation, and the cross-disciplinary collaboration necessary to translate microbiome insights into clinical practice.

Conclusion: a line drawn under conflicting claims

By applying an exceptionally rigorous analysis pipeline to one of the world’s largest cancer genome datasets, the researchers provide a clearer map of where microbiomes exist in cancer. Digestive tract cancers appear to host microbial life within tumours, while other cancers do not show a microbiome distinguishable from background once contamination is carefully controlled. The work highlights the non-trivial problem of contamination, demonstrates a path to robust detection, and offers data and tools that empower other scientists to test, refine, and extend these findings. The Conversation emphasizes that as research continues, the field should concentrate on the digestive tract cancers where microbial communities are evident and continue to refine methods to distinguish true biological signal from laboratory artefact.