Abstract

Objective: This study investigated spatial heterogeneity and immune organization in triple-negative breast carcinoma using single-cell-referenced spatial transcriptomic deconvolution. Methods: In tumour samples profiled by both single-cell RNA sequencing and spatial transcriptomics, each spatial spot was deconvolved by non-negative least squares against the single-cell reference, and the recovered composition was analysed by spatial autocorrelation, niche clustering and co-localization. Results: Deconvolution accurately recovered major tumour and stromal populations, with correlations of 0.981 for malignant cells and 0.952 for cancer-associated fibroblasts. All evaluated cell types exhibited significant spatial autocorrelation. Malignant cells showed the strongest clustering (Moran’s I = 0.927), followed by B cells (I = 0.911), endothelial cells (I = 0.869), and cancer-associated fibroblasts (I = 0.817). Five spatial niches were identified, representing distinct malignant, stromal, lymphoid, vascular, and peritumoural compartments. CD8 T-cell abundance was lowest within the malignant core (8.1%), peaked at the peritumoural margin (14.1%), and declined toward the tissue periphery (3.5%). Conclusion: The triple-negative breast tumour microenvironment displayed pronounced spatial organization, characterized by distinct cellular niches and preferential accumulation of cytotoxic T-cells at the peritumoural interface rather than within malignant regions.

Keywords: Triple-negative breast cancer, Spatial transcriptomics, Single-cell RNA sequencing, Deconvolution, Tumour microenvironment, Immune exclusion, Cancer associated fibroblasts, Spatial niches