Related Experiment Video
Updated: Dec 7, 2025
06:53
Detection and Monitoring of Tumor Associated Circulating DNA in Patient Biofluids
Published on: June 8, 2019
9.0K
Machine Learning-Assisted Evaluation of Circulating DNA Quantitative Analysis for Cancer Screening
Rita Tanos1,2,3, Guillaume Tosato1,2,3,4, Amaelle Otandault1,2,3
1IRCM Institut de Recherche en Cancérologie de Montpellier INSERM U1194 Montpellier F-34090 France.
Advanced Science (Weinheim, Baden-Wurttemberg, Germany)
|October 1, 2020
Summary
This study introduces a novel method for cancer screening using quantitative and structural features of cell-free DNA (cfDNA). Machine learning models show high accuracy in detecting early-stage colorectal cancer.
Area of Science:
- Biochemistry
- Genomics
- Computational Biology
Background:
- Circulating cell-free DNA (cfDNA) analysis is emerging for cancer detection.
- Current methods often focus on genetic and epigenetic alterations.
- An alternative approach using cfDNA's physical characteristics is explored.
Purpose of the Study:
- To develop and validate a novel cancer screening strategy using cfDNA quantitative and structural features.
- To assess the performance of machine learning models in classifying cancer patients versus healthy individuals based on these features.
- To demonstrate the potential for early-stage cancer detection.
Main Methods:
- Demonstrated cfDNA quantitative and structural parameters in cell culture, murine, and human plasma models.
- Evaluated these parameters in a large retrospective cohort (289 healthy, 983 cancer patients).
- Employed a machine learning decision tree model for classification after age resampling.
Main Results:
- The developed model achieved high performance in detecting and classifying healthy individuals and cancer patients.
- Unprecedented performance was observed for early-stage colorectal cancer (Stage 0, I, II) with 0.89 specificity and 0.72 sensitivity.
- The study confirmed the potential of quantitative and structural cfDNA biomarkers.
Conclusions:
- A novel, efficient strategy for cancer screening using cfDNA quantitative and structural biomarkers combined with machine learning is established.
- This approach shows significant promise for early cancer detection, particularly for colorectal cancer.
- Future optimization may involve integrating these biomarkers with fragmentomics, methylation, and genetic alteration detection for enhanced classification rates.
