Resolving tumor microenvironment heterogeneity to forecast immunotherapy response in triple-negative breast cancer through multi-scale analysis.

Background: Immunotherapy has been used in the clinical management of TNBC. While BRCA1 mutations are associated with immunotherapy response, the therapeutic outcomes in TNBC patients are not promising. Methods: This study integrated spatial, single-cell, and bulk RNA-seq data to explore the role of BRCA1 in reshaping the TNBC microenvironment. Through multi-scale analysis, phenotype changes and potential biomarkers in cancer-associated fibroblasts (CAF) were identified. To validate these findings at the protein level, we employed high-resolution, label-free proteomics sequencing in our in-house cohort, providing critical real-world validation. A predictive system for response to ICIs was constructed through the step-by-step machine learning pipeline. Results: Compared to BRCA1 mutant patients, BRCA1 wild-type patients experienced increased T-cell exhaustion and dendritic cell tolerance. We identified a MEG3+ pre-CAF subgroup via pseudo-time analysis. Moreover, ISG15 may serve as an immunoregulatory biomarker, and the proposed predictive model demonstrated potential in forecasting immunotherapy response, although further validation is needed. Conclusions: This study highlighted the cellular heterogeneity of TNBC and identified ISG15 as a candidate biomarker potentially associated with treatment response. The ISG15-based predictive system might provide a robust framework for predicting ICI response.