Related Experiment Video
Updated: May 21, 2025
08:40
Methyl-binding DNA capture Sequencing for Patient Tissues
Published on: October 31, 2016
8.6K
Exploring potential methylation markers for ovarian cancer from cervical scraping samples
Ju-Yin Lien1, Lu Ann Hii1, Po-Hsuan Su2
1Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taipei, Taiwan.
Human Genomics
|May 17, 2025
Summary
This study developed a machine learning model using DNA methylation from Pap test cervical scrapings to detect ovarian cancer. The model shows promise for early ovarian cancer detection and diagnosis using a non-invasive sample type.
Area of Science:
- Gynecology
- Oncology
- Genomics
- Machine Learning
Background:
- Ovarian cancer has a high mortality rate, necessitating early detection for improved survival.
- Current diagnostic methods lack specificity and sensitivity, and are often invasive.
- The Papanicolaou (Pap) test shows potential for ovarian cancer detection via tumor DNA in cervical scrapings.
Purpose of the Study:
- To develop a machine learning model for ovarian cancer stratification using DNA methylation data.
- To analyze cervical scraping samples collected during Pap tests for ovarian cancer biomarkers.
- To establish a non-invasive method for early ovarian cancer detection.
Main Methods:
- Collected 160 cervical scraping samples (95 normal, 37 benign, 28 malignant).
- Generated DNA methylation data using the Illumina Infinium MethylationEPIC BeadChip array.
- Developed a two-step machine learning model involving Principal Component Analysis (PCA) and gradient boosting for classification.
Main Results:
- The two-step model achieved 0.88 accuracy and 0.86 F1-score on test data.
- Differential methylation analysis identified potential gene functions related to olfactory transduction (normal vs. tumor) and immune regulation (benign vs. malignant).
Conclusions:
- The developed two-step model demonstrates potential for ovarian cancer prediction.
- Cervical scrapings collected via Pap tests are a viable alternative sample source for screening.
- DNA methylation analysis in cervical samples offers a promising non-invasive approach for ovarian cancer diagnosis.
