Reliable blood-based tests for identifying early-stage breast cancer remain elusive. Employing single-cell transcriptomic sequencing analysis, we illustrate a close correlation between nucleotide metabolism in the breast cancer and activation of regulatory T cells (Tregs) in the tumor microenvironment, which shows distinctions between subtypes of patients with triple-negative breast cancer (TNBC) and non-TNBC, and is likely to impact cancer prognosis through the A2AR-Treg pathway. Combining machine learning with absolute quantitative metabolomics, we have established an effective approach to the early detection of breast cancer, utilizing a four-metabolite panel including inosine and uridine. This metabolomics study, involving 1111 participants, demonstrates high accuracy across the training, test, and independent validation cohorts. Inosine and uridine prove predictive of the response to neoadjuvant chemotherapy (NAC) in patients with TNBC. This study deepens our understanding of nucleotide metabolism in breast cancer development and introduces a promising non-invasive method for early breast cancer detection and predicting NAC response in patients with TNBC.
Keywords: Cancer; Machine learning; Metabolomics.
© 2024 The Author(s).