A predictive model for neoadjuvant therapy response in breast cancer.

Neoadjuvant therapy is a standard treatment for breast cancer, but its effectiveness varies among patients. This highlights the importance of developing accurate predictive models. Our study uses metabolomics and machine learning to predict the response to neoadjuvant therapy in breast cancer patients.

Objective: To develop and validate predictive models using machine learning and circulating metabolites for forecasting responses to neoadjuvant therapy among breast cancer patients, enhancing personalized treatment strategies.

Methods: Based on pathological analysis after neoadjuvant chemotherapy and surgery, this retrospective study analyzed 30 young women breast cancer patients from a single institution, categorized as responders or non-responders. Utilizing liquid chromatography-tandem mass spectrometry, we investigated the plasma metabolome, explicitly targeting 40 metabolites, to identify relevant biomarkers linked to therapy response, using machine learning to generate a predictive model and validate the results.

Results: Eighteen significant biomarkers were identified, including specific acylcarnitines and amino acids. The most effective predictive model demonstrated a remarkable accuracy of 90.7% and an Area Under the Curve (AUC) of 0.999 at 95% confidence, illustrating its potential utility as a web-based application for future patient management. This model's reliability underscores the significant role of circulating metabolites in predicting therapy outcomes.

Conclusion: Our study's findings highlight the crucial role of metabolomics in advancing personalized medicine for breast cancer treatment by effectively identifying metabolite biomarkers correlated with neoadjuvant therapy response. This approach signifies a critical step towards tailoring treatment plans based on individual metabolic profiles, ultimately improving patient outcomes in breast cancer care.