Prostate cancer detection with blood serum autofluorescence: Comparison to non-optical methods

Abstract

Blood serum fluorescence spectroscopy provides a rapid and facile method for assessing protein conformational states and metabolic shifts, rendering it highly valuable for the diagnostic investigation of pathologies. Here, we demonstrate the potential of blood serum fluorescence spectroscopy for prostate cancer diagnosis. Current diagnostic modalities for prostate cancer encompass blood tests for prostate-specific antigen level, digital rectal examination, ultrasound and MRI assay. Despite their widespread utilization, these established techniques exhibit limitations in specificity, resulting in an increased number of unnecessary biopsies. Based on the analysis of blood serum fluorescence emission spectra, we identified a novel potential marker - asymmetry of fluorescence spectra at 350 nm excitation, Asym350, for prostate cancer, which demonstrated statistically significant discriminatory power in identifying prostate cancer patients (p < $5\times {10}^4$). This fluorescence-derived marker was integrated into a classification model alongside PSA level and PI-RADS score. By using cross-validation to evaluate the performance of classification model across various feature sets, we achieved the highest F1 score of 0.91 when utilizing feature set of PI-RADS and Asym350. These findings underscore the capabilities of blood serum fluorescence spectroscopy for prostate cancer diagnostics and raise the crucial question of the feasibility of its translation into clinical application.