Answers are grounded in this paper and cite their sources.
Objectives: Neoadjuvant chemotherapy (NAC) improves the prognosis of early and locally advanced breast cancer, yet pathological complete response ( pCR) is difficult to achieve in locally advanced cases, restricting therapeutic efficacy. This study aimed to explore whether spatial tumor microenvironment (TME) features of pretreatment primary tumors (PT) and paired lymph node metastases (LN) predict NAC response in breast cancer. Materials and Methods: Multiplex immunofluorescence was used to analyze the spatial TME phenotypes of pretreatment PT and paired LN samples from breast cancer patients. Differences in TME cell abundance and spatial proximity to tumor cells were compared between patients who achieved pCR and those who did not (non-pCR). Machine learning models were constructed based on single-site (PT-only or LN-only) or dual-site (PT+LN) TME features to predict NAC response. Results: Across both PT and LN lesions, the non-pCR group exhibited enriched metabolic support cells (adipocytes) and immunosuppressive cells (M2 macrophages) with tighter spatial proximity to tumor cells, accompanied by the exclusion of antitumor effector cells (cytotoxic T lymphocytes, B cells, non-M2 macrophages, and dendritic cells). Distinct site-specific TME patterns were identified: in PT, non-pCR showed increased angiogenesis, alongside higher density and closer spatial contact of potentially exhausted T cells and regulatory T cells with tumor cells, while non-pCR LN were characterized by elevated neutrophil tumor proximity and reduced B cell-tumor contact. Machine learning analysis revealed that the dual-site (PT+LN) TME model showed higher predictive performance compared with single-site models. Conclusion: Spatial TME signatures of PT and LN may serve as predictors of NAC response. Integrating dual-site TME features improves predictive accuracy, and the LN microenvironment provides incremental predictive value beyond primary tumor profiling.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: