Machine Learning-Enhanced Analysis of Exosomal Surface Sialic Acid Using Surface-Enhanced Raman Spectroscopy for Ovarian Cancer Diagnosis and Therapeutic Monitoring.

Currently, the absence of ovarian cancer (OC)-specific biomarkers impedes the development of precise noninvasive diagnostic and monitoring strategies. Exosomal surface sialic acid (SA), a key mediator of intercellular communication and disease progression, emerges as a promising biomarker, though its role in OC remains unclear. Conventional exosome isolation and detection methods exhibit limited clinical utility. Herein, we developed a CD63 aptamer-functionalized gold array chip integrated with a surface-enhanced Raman scattering (SERS) nanosensor for sensitive SA analysis. The chip efficiently isolated exosomes from clinical serum, while the nanosensor selectively bound exosomal SA via molecular recognition, thereby altering the SERS intensity ratio of the nanosensor. More importantly, machine learning can discern SA signatures from SERS spectra, achieving 93% accuracy in OC diagnosis. The longitudinal monitoring of SA throughout the entire treatment period (preoperative, postoperative, and chemotherapy) revealed a potential correlation with treatment response as indicated by clinical markers (CA125, HE4), demonstrating the utility of exosomal SA in precision treatment evaluation. This provides a powerful tool for the diagnosis and treatment monitoring of OC and plays a critical role in precision medicine.