Computational stratification and engineering framework of the PDCD1/CD2 immune axis in triple-negative breast cancer

Purpose Triple-negative breast cancer (TNBC) is a high-risk molecular subtype defined by absence of estrogen receptors, progesterone receptors, and human epidermal growth factor receptor 2 (HER2) overexpression. Immune checkpoint blockade (ICB) benefits only 20% to 40% of patients, with T cell exhaustion driven by chronic programmed death receptor 1 (PD-1, PDCD1) inhibitory signalling being the dominant barrier. This work presents a computational pipeline that converts single-cell immune phenotypes into per-patient synthetic immune-cell engineering recommendations, filling the gap for quantitative patient-level ligand design parameters. Methods The pipeline was applied to the GSE176078 single-cell RNA sequencing (scRNA-seq) atlas (100,064 cells, 26 treatment-naive TNBC patients). After quality control, normalisation, and Leiden clustering, T cells were extracted and scored for exhaustion and cytotoxicity using defined gene-set modules. The PDCD1/CD2 ratio was computed per cell and aggregated per patient. Cross-modal validation used TCGA-BRCA bulk RNA-seq through univariate Cox regression and Kaplan–Meier analysis. A targeted ligand-receptor proxy screen was performed across five receptor-ligand pairs and a rule-based DesignPriorityScore was assessed by bootstrap resampling ( n = 200). Results Leiden clustering was validated at adjusted rand index (ARI) 0.288 to 0.311 and normalised mutual information (NMI) 0.616 to 0.671. Three patient phenotype groups were identified based on exhaustion burden and PDCD1/CD2 imbalance. The bulk PDCD1/CD2 ratio showed an exploratory association with overall survival in TCGA-BRCA (HR=0.47, 95%CI 0.28–0.79; P < 0.005), reflecting immune infiltration rather than per-cell exhaustion state. Patient rankings were stable across bootstrap resamples (mean Spearman ρ > 0.85; top-quartile retention > 90%). Conclusion This pipeline shows that per-patient PDCD1/CD2 ratios derived from scRNA-seq can be translated into ranked synthetic ligand engineering priorities, offering a prototype framework for single-cell-informed synthetic immunology design in TNBC immunotherapy.