Urine Analysed by FTIR, Chemometrics and Machine Learning Methods in Determination Spectroscopy Marker of Prostate Cancer in Urine

Abstract

Prostate-specific antigen (PSA) is the most commonly used marker of prostate cancer. However, nearly 25% of men with elevated PSA levels do not have cancer and nearly 20% of patients with prostate cancer have normal serum PSA levels. Therefore, in this study, Fourier transform infrared (FTIR) spectroscopy was investigated as a new tool for detection of prostate cancer from urine. Obtained results showed higher levels of glucose, urea and creatinine in urine collected from patients with prostate cancer than that in control. Principal component analysis (PCA) was not noticed possibility of differentiation urine collected from healthy and nonhealthy patients. However, machine learning algorithms showed 0.90 accuracy and precision of FTIR in detection of prostate cancer from urine. We showed that wavenumbers at 1614 cm−1 and 2972 cm−1 were candidates for prostate cancer spectroscopy markers. Importantly, these FTIR markers correlated with Gleason score, PSA and mpMRI PI-RADS category.