Single-cell transcriptomics reveals cellular heterogeneity and neoadjuvant chemotherapy response signatures in triple-negative breast cancer

Abstract

Background Triple-negative breast cancer (TNBC) shows marked intratumoral heterogeneity and variable responses to neoadjuvant chemotherapy (NAC), but the cellular determinants of treatment response remain incompletely defined. Methods We analyzed the scRNA-seq dataset GSE161529 to characterize TNBC cellular states and used bulk cohorts, including GSE25066 and GSE58812, for NAC response-related signature evaluation and clinical validation. Cell types and subtypes were identified by t-SNE clustering. A permutation-based NACR score estimated chemotherapy-response potential. Differentiation hierarchies were reconstructed using CytoTRACE 2 and Monocle 2, and cell-cell communication was inferred with CellChat. Prognostic value was evaluated by Kaplan-Meier analysis, drug sensitivity was predicted using pRRophetic/GDSC, and in vitro validation used MDA-MB-231-derived C2-like sensitive and C4-like resistant TNBC models. Results Nine major cell populations were identified, with cancer cells (49.3%) showing the highest NACR scores. Four cancer cell subtypes were delineated: C2 was enriched in predicted NACR-high groups, whereas C3/C4 correlated with low predicted response. Three tumor-associated macrophage (TAM) subtypes were identified, with TAMs + SLPI associated with high predicted NACR and TAMs + CXCL9 with low predicted NACR. Pseudotime analysis revealed increasing NACR scores along the cancer cell trajectory (C4 to C2) and decreasing scores along the TAM trajectory. Predicted NACR-high tumors showed dominant CAF-cancer cell interactions, whereas predicted NACR-low tumors showed endothelial cell-CAF/TAM communication via PDGFB-PDGFRB and ICAM2-ITGAM/ITGB2. The NACR signature was associated with overall survival (log-rank p = 0.034). In vitro experiments confirmed a 12.81-fold IC50 difference between C2-like and C4-like models and validated subtype markers. Conclusions This integrative analysis identifies response-associated cancer cell and TAM states and nominates PDGFB-PDGFRB signaling as a candidate resistance-associated pathway, providing potential biomarkers and therapeutic targets for precision TNBC therapy.