New algorithm to predict colorectal cancer based on fecal volatile organic compounds profile

Area of Science:

• Oncology
• Analytical Chemistry
• Biotechnology

Background:

• Colorectal cancer (CRC) diagnosis relies on invasive procedures.
• Early detection significantly improves patient outcomes.
• Advanced analytical techniques offer potential for non-invasive diagnostics.

Purpose of the Study:

• To develop and validate an algorithm for CRC diagnostics using fecal samples.
• To differentiate between colorectal cancer, adenomas, and healthy controls.
• To identify specific compound patterns indicative of different health states.

Main Methods:

• Utilized thermal-desorption-gas chromatography-mass spectrometry (TD-GC-MS) for fecal sample analysis.
• Developed a comprehensive algorithm analyzing the entire spectral range.
• Optimized the algorithm for sensitivity and specificity in diagnostic classification.

Main Results:

• Optimized algorithm achieved 100% sensitivity and specificity in initial testing.
• Validation phase showed varying performance (sensitivity 68-74%, specificity 52-58%, accuracy 62-66%).
• Algorithm demonstrated potential for polyp sample analysis even with limited training data.

Conclusions:

• The developed algorithm represents a significant advancement in non-invasive CRC diagnostics.
• Compound pattern analysis provides interpretability for diagnostic predictions.
• This approach holds promise for enhancing early detection and precision in colorectal cancer diagnosis.