Scientists have developed a plasma test that can detect colorectal cancer early by profiling microbiome cfRNA methylation. Their test offers higher accuracy than current noninvasive methods.
Image Credit: Elif Bayraktar/Shutterstock.com
In a study published in Nature Biotechnology, researchers have unveiled a promising new method for noninvasive colorectal cancer (CRC) screening that could improve early diagnosis and patient outcomes.
Current tools, such as imaging and biopsies, often miss early-stage detection, while liquid biopsies based on cell-free DNA are limited by low sensitivity and scarcity of tumor DNA in blood. Instead, the research team turned to cell-free RNA (cfRNA), which captures a broader range of molecular information, including gene transcripts and their chemical modifications.
They focused on cfRNA from the body’s microbiome, the vast community of bacteria and viruses living in and on the human body. These microbial populations change during cancer development, and fluctuations in the biome can indicate cancerous growth.
The study aimed to determine whether specific methylation patterns in microbiome-derived cfRNA could serve as reliable early indicators of colorectal cancer.
How the Test Works
Plasma samples were collected from individuals with CRC and from noncancer controls and processed using rigorous protocols to preserve cfRNA integrity. After multiple high-speed centrifugation steps to remove cellular debris, samples were stored at ultra-low temperatures.
cfRNA was extracted using a commercial low-input kit capable of recovering small RNAs, including transfer RNAs (tRNAs), which are known to exhibit extensive methylation.
To detect these methylation modifications, the researchers developed a sequencing strategy called LIME-seq. This method applies enzymatic treatments that reveal methylated versus unmethylated nucleotides, enabling direct detection of methylation signatures in plasma-derived small RNAs.
Sequencing data were then analyzed using bioinformatic pipelines, including Kraken2, to identify microbial species contributing cfRNA and to map methylation sites with nucleotide-level resolution.
Machine learning models, specifically support vector machines, were trained on these methylation profiles from a discovery cohort and validated using leave-one-out cross-validation to assess robustness.
The team also compared methylation-based classification to other microbial biomarkers, such as microbial abundance and mutational signatures, to evaluate relative diagnostic performance.
Impressive Results
The test identified a set of methylation sites consistently altered in colorectal cancer patients, pointing to disrupted microbiome activity associated with tumor growth. These changes were observed across a variety of microbial species, suggesting widespread changes in the microbiome.
The diagnostic models based on methylation patterns achieved an area under the curve of around 0.98, indicating excellent accuracy. Crucially, the method proved effective at detecting early-stage disease, including precancerous adenomas and stage I cancers—areas where current screening methods often fall short.
When the team combined methylation patterns with mutational signatures also present in cfRNA, the test’s performance improved even further. The researchers say the approach is less affected by the confounding factors that limit DNA-based tests and offers real-time insight into the microbiome’s functional state.
Download your PDF copy now!
Looking Ahead
The study's results show a significant step towards more sensitive and specific noninvasive screening for colorectal cancer. By capturing dynamic molecular changes in the microbiome, cfRNA methylation profiling could enhance early detection and guide risk assessment.
It also highlights the potential of applying this approach to other diseases linked to microbiome dysfunction.
Journal Reference
Ju CW., Lyu R., et al. (2025). Modifications of microbiome-derived cell-free RNA in plasma discriminates colorectal cancer samples. Nature Biotechnology. DOI: 10.1038/s41587-025-02731-8, https://www.nature.com/articles/s41587-025-02731-8