JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Derivation and validation of lifestyle-based and microbiota-based models for colorectal adenoma risk evaluation and self-prediction

  • 0Division of Gastroenterology and Hepatology, NHC Key Laboratory of Digestive Diseases, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China.
Bmj Open Gastroenterology +

|

April 2, 2025

|

April 2, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

A new non-invasive model using lifestyle and gut bacteria predicts colorectal adenoma (CRA) risk. This tool aids in early colorectal cancer (CRC) prevention by improving screening efficacy and identifying personalized screening ages.

Area Of Science

  • Gastroenterology and Oncology
  • Microbiome Research
  • Preventive Medicine

Background

  • Early detection of colorectal adenoma (CRA) is crucial for colorectal cancer (CRC) prevention.
  • Current screening methods can be invasive and may have limitations in efficacy.
  • Non-invasive strategies are needed to improve population-level screening.

Purpose Of The Study

  • To construct a non-invasive prediction model for colorectal adenoma (CRA) risk stratification.
  • To enhance the efficacy of CRA screening for colorectal cancer (CRC) prevention.
  • To determine optimal initial colonoscopy screening ages for different risk groups.

Main Methods

  • Integrated three cohorts (9747 participants) undergoing colonoscopy.
  • Utilized lifestyle information, fecal samples, 16S rRNA sequencing, and quantitative real-time PCR to identify CRA-associated bacteria.
  • Developed prediction models using lifestyle and gut microbiota data, validated across cohorts, and analyzed age-specific CRA incidence rates.

Main Results

  • A multivariable logistic regression model incorporating 14 variables demonstrated robust CRA prediction (c-statistic=0.79).
  • Machine learning models showed comparable performance (random forest: 0.78, gradient boosting: 0.78).
  • Inclusion of specific bacteria (Fusobacterium nucleatum, pks+ E. coli) improved model performance (c-statistic=0.84-0.86); recommended screening ages: 42 (high-risk) vs. 53 (low-risk).

Conclusions

  • An integrated mathematical model combining lifestyle and gut microbial signatures offers an effective non-invasive strategy for CRA risk stratification.
  • A machine learning-assisted prediction application facilitates cost-effective, population-level screening.
  • This approach optimizes colorectal cancer (CRC) prevention protocols through improved screening implementation.

Keywords: