A Novel Method for Colorectal Cancer Screening Based on Circulating Tumor Cells and Machine Learning

Area of Science:

• Oncology
• Biotechnology
• Computational Biology

Background:

• Colorectal cancer (CRC) is a leading cause of cancer mortality, often diagnosed at late stages.
• Current screening methods like colonoscopy are invasive and burdensome.
• Early detection is crucial for improving patient outcomes and survival rates.

Purpose of the Study:

• To develop and validate a machine learning classifier for distinguishing colorectal cancer from non-cancerous samples.
• To assess the efficacy of enumerating circulating tumor cells (CTCs) using flow cytometry for cancer detection.
• To explore the potential of a non-invasive blood-based screening tool for CRC.

Main Methods:

• Enumeration of circulating tumor cells (CTCs) via flow cytometry.
• Development of a Support Vector Machine (SVM) classifier trained on CTC counts.
• Validation of the SVM classifier on a blind dataset and comparison with other machine learning models using SMOTE for over-sampling.

Main Results:

• The SVM classifier demonstrated high performance with 90.0% accuracy, 80.0% sensitivity, 100.0% specificity, 100.0% precision, and an AUC of 0.98 on blind samples.
• SVM outperformed other machine learning models in discriminating between healthy and colorectal cancer patients based on validation accuracy.
• The study confirmed that CTC enumeration by flow cytometry provides significant data for machine learning-based cancer discrimination.

Conclusions:

• Circulating tumor cells (CTCs) enumerated by flow cytometry are valuable biomarkers for CRC detection.
• Machine learning algorithms, particularly SVM, can effectively utilize CTC data to differentiate cancer patients.
• This approach holds promise for developing a simple, rapid, and non-invasive screening tool for colorectal cancer.