What question did this study set out to answer?

This study aims to establish a predictive model based on microbial signatures for colorectal adenoma detection.

June 20, 2026Open Access

Unveiling the unique gut microbial signatures in colorectal adenomas: establishment and validation of a cross-kingdom microbiome predictive model

Key Points

This study aims to establish a predictive model based on microbial signatures for colorectal adenoma detection.
Collected 296 fecal samples from multiple centers between January 2019 and January 2024.
Performed metagenomic analysis, including shotgun sequencing and principal coordinate analysis.
Developed a random forest model for distinguishing adenoma patients from healthy individuals with internal and external validation.
Significant differences in microbial abundance were found in adenoma patients, including specific bacteria, archaea, fungi, and viruses.
The prediction model achieved a mean AUC of 0.80 ± 0.05 for internal validation and an AUC of 0.75 in external validation.
Complex network relationships among microorganisms were identified through Spearman correlation analysis.

Abstract

Background Colorectal adenoma (CA), the main precancerous lesion of colorectal cancer (CRC), originates in approximately 85–90% of CRC cases. With increasing demands for early diagnosis and treatment, gut microbiome research has become a forefront area. While numerous studies have shown that gut bacteria are closely related to the development of colorectal adenomas and cancer, research on viruses, archaea, and fungi is limited. Methods From January 2019 to January 2024, this study collected 296 fecal samples from multiple centers and performed metagenomic analysis using shotgun sequencing. Principal coordinate analysis (PCoA) was conducted based on Bray-Curtis distance at the species level, α-diversity was calculated, and LEfSe analysis identified differential microorganisms. A random forest model was developed to distinguish adenoma patients from healthy individuals, with performance evaluated through internal validation using Bootstrap sampling and external validation with an independent cohort. Findings Significant differences in the relative abundance of certain bacteria (e. g. , Phocaeicolaᵥulgatus and Prevotellacopri), fungi (Candidaₐlbicans), archaea (Methanobrevibacterₒralis), and viruses (Streptococcus satellite phage Javan301) were observed in adenoma patients. Spearman correlation analysis revealed complex network relationships among these microorganisms. The prediction model achieved a mean AUC of 0. 80 ± 0. 05 and an external validation AUC of 0. 75, demonstrating stability and generalizability. Conclusion This study shows significant cross-kingdom microbial signatures in colorectal adenoma patients, providing potential for developing new preventive and therapeutic methods. The predictive model, based on these differential microorganisms, exhibits robust and promising classification performance, offering potential for early adenoma detection.

Bookmark

View Full Paper