For years, reports have suggested that microbial communities might be linked to nearly every type of cancer. A new analysis from Johns Hopkins Medicine challenges that view, finding only modest amounts of microbial DNA in tumor sequencing data and raising concerns about contamination in earlier studies.
Published in Science Translational Medicine, the study examined 5,734 samples from 25 cancer types available in The Cancer Genome Atlas. After removing human DNA using Telomere-to-Telomere and Genome Reference Consortium assemblies, the team compared the remaining reads against a reference containing more than 50,000 microbial genomes. On average, only 0.57% of reads from solid tumors and 0.73% from blood cancers mapped to microbial sequences.
“Over time, we see a more complete picture of new research, and in this case, we did not find any associations between microbiomes and many types of cancer,” said Steven Salzberg, Bloomberg Distinguished Professor at Johns Hopkins. He emphasized that rigorous contamination controls were essential to avoid false positives, noting that sequencing labs are prone to environmental DNA, including common contaminants such as Saccharomyces cerevisiae.
The researchers contrasted their results with prior studies. A retracted Nature paper reported microbial read counts dozens to thousands of times higher, while a 2022 Cell paper suggested fungal DNA amounts that the Johns Hopkins team attributes largely to contamination. Still, the group confirmed the presence of established cancer-associated microbes, including HPV, Helicobacter pylori, Fusobacterium nucleatum, and Bacteroides fragilis.
Salzberg said documenting these discrepancies is “especially important” as researchers consider microbiome-based diagnostics for cancer. The full dataset and analysis pipeline are available in the study’s supplementary materials.
Publication Details
Yuchen Ge et al., Comprehensive analysis of microbial content in whole-genome sequencing samples from The Cancer Genome Atlas project. Sci. Transl. Med. 17, eads6335 (2025). DOI:10.1126/scitranslmed.ads6335
Published in Science Translational Medicine, the study examined 5,734 samples from 25 cancer types available in The Cancer Genome Atlas. After removing human DNA using Telomere-to-Telomere and Genome Reference Consortium assemblies, the team compared the remaining reads against a reference containing more than 50,000 microbial genomes. On average, only 0.57% of reads from solid tumors and 0.73% from blood cancers mapped to microbial sequences.
“Over time, we see a more complete picture of new research, and in this case, we did not find any associations between microbiomes and many types of cancer,” said Steven Salzberg, Bloomberg Distinguished Professor at Johns Hopkins. He emphasized that rigorous contamination controls were essential to avoid false positives, noting that sequencing labs are prone to environmental DNA, including common contaminants such as Saccharomyces cerevisiae.
The researchers contrasted their results with prior studies. A retracted Nature paper reported microbial read counts dozens to thousands of times higher, while a 2022 Cell paper suggested fungal DNA amounts that the Johns Hopkins team attributes largely to contamination. Still, the group confirmed the presence of established cancer-associated microbes, including HPV, Helicobacter pylori, Fusobacterium nucleatum, and Bacteroides fragilis.
Salzberg said documenting these discrepancies is “especially important” as researchers consider microbiome-based diagnostics for cancer. The full dataset and analysis pipeline are available in the study’s supplementary materials.
Publication Details
Yuchen Ge et al., Comprehensive analysis of microbial content in whole-genome sequencing samples from The Cancer Genome Atlas project. Sci. Transl. Med. 17, eads6335 (2025). DOI:10.1126/scitranslmed.ads6335