Urinary volatile organic compounds (VOCs) based prostate cancer diagnosis via high-dimensional classification

Area of Science:

• Oncology
• Biomarker Discovery
• Computational Biology

Background:

• Early prostate cancer (PCa) detection is vital for treatment success.
• Current methods like PSA testing and DRE have diagnostic limitations.
• Urinary volatile organic compounds (VOCs) are emerging as potential PCa biomarkers.

Purpose of the Study:

• To develop and evaluate a diagnostic model for prostate cancer using urinary VOCs.
• To address the high-dimensional classification challenge posed by numerous VOCs.
• To compare the diagnostic performance of VOC-based models against traditional methods like PSA testing.

Main Methods:

• Collected urine samples from 337 PCa-positive and 233 PCa-negative patients.
• Employed high-dimensional classification techniques to analyze VOC data.
• Utilized regularized logistic regression, specifically with the SCAD penalty, for model development.

Main Results:

• The regularized logistic regression model with SCAD penalty demonstrated strong performance.
• The final VOC-based model achieved an Area Under the ROC Curve (AUC) of 0.748.
• This significantly outperformed the AUC of 0.540 obtained from PSA testing.

Conclusions:

• Urinary VOCs represent a promising, non-invasive biomarker for prostate cancer screening.
• Regularized logistic regression, particularly with SCAD penalty, is effective for analyzing high-dimensional VOC data.
• VOC-based diagnosis offers a potentially more accurate and clinically feasible approach for PCa detection compared to current standards.