Breast Cancer Subtyping by Surface-Enhanced Raman Scattering (SERS) Signature of Plasma-Derived Extracellular Vesicles

Abstract Breast cancer subtype classification has critical implications for patients’ treatment selection and disease prognosis. Small extracellular vesicles (sEVs) have emerged as promising biomarkers for liquid biopsy due to their ability to reflect the molecular characteristics of their parent cells and are found in high concentration in biological fluids, such as plasma and urine. However, accurately distinguishing breast cancer subtypes using sEVs remains challenging because of the subtle biochemical differences among heterogeneous sEV populations. In this study, label-free surface-enhanced Raman scattering (SERS) was employed to analyze plasma-derived sEVs as a comprehensive molecular profiling approach, capable of capturing biochemical information without prior target selection or labelling. To extract discriminative features from complex spectral datasets among sEV populations, label-free SERS was combined with a support vector machine (SVM) classifier, establishing a precise, data-driven framework for molecular subtyping. The integrated SERS-SVM platform successfully differentiated human epidermal growth factor receptor 2 negative (HER-2 neg ) breast cancer, HER-2 positive (HER-2 pos ) breast cancer, and healthy controls, highlighting its potential as a rapid, non-invasive, and high-throughput liquid biopsy approach for breast cancer diagnosis and molecular subtyping. In our study, the SVM model achieved an overall accuracy of 95% in internal train–test validation and 85% in an independent validation using a separate set of clinical samples. Importantly, the independent test set showed 100% sensitivity for both HER-2 neg and HER-2 pos breast cancer groups, indicating that the model correctly identified all breast cancer subtype samples in the validation cohort.