Raman Spectroscopy and Exosome-Based Machine Learning Predicts the Efficacy of Neoadjuvant Therapy for HER2-Positive Breast Cancer

Abstract

Early prediction of the neoadjuvant therapy efficacy for HER2-positive breast cancer is crucial for personalizing treatment and enhancing patient outcomes. Exosomes, which play a role in tumor development and treatment response, are emerging as potential biomarkers for cancer diagnosis and efficacy prediction. Despite their promise, current exosome detection and isolation methods are cumbersome and time-consuming and often yield limited purity and quantity. In this study, we employed Raman spectroscopy to analyze the molecular changes in exosomes from the sera of HER2-positive breast cancer patients before and after two cycles of neoadjuvant therapy. Utilizing machine learning techniques (PCA, LDA, and SVM), we developed a predictive model with an AUC value exceeding 0.89. Additionally, we introduced an innovative HER2-positive exosome capture and detection system, termed Magnetic beads@HER2-Exos@HER2-SERS detection nanoprobes (HER2-MEDN). This system enabled us to efficiently extract and analyze HER2-positive exosomes, refining our predictive model to achieve an accuracy greater than 0.94. Our study has demonstrated the potential of the HER2-MEDN system in accurately predicting early treatment response, offering novel insights and methodologies for assessing the efficacy of neoadjuvant therapy in HER2-positive breast cancer.