The immunosuppressive tumor microenvironment (TME) is a key driver of prostate cancer (PCa) progression, yet the dynamic interplay between stromal and immune cells remains poorly characterized. Through integrative analysis of single-cell transcriptomes and clinical cohorts, we identified a coordinated stromal-immune axis involving tumor-associated macrophages(TAM) and cancer-associated fibroblasts (CAFs) that orchestrates TME remodeling.

Single-cell RNA sequencing (scRNA-seq) was performed on 48 PCa and 25 adjacent normal tissues. Clinical correlations were validated in the TCGA cohort and two independent PCa cohorts. Intercellular communication (CellChat) and spatial co-localization (multiplex immunofluorescence, mIF) were employed to decode intercellular communication. A machine learning-based composite risk score integrating SPP1+ macrophage and FAP+ fibroblast features was developed to predict progression-free survival (PFS). Xenograft model was utilized to assess the therapeutic efficacy of dual-targeting strategy.

scRNA-seq identified SPP1+ macrophages as a predominant M2-polarized subset enriched in tumors versus normal tissues. Their abundance positively correlated with FAP+ fibroblast (r=0.62, p<0.001) and inversely with cytotoxic T lymphocytes (CTLs, r=-0.48, p=0.007). CellChat analysis predicted CD99-CD99L2 ligand-receptor signaling as a key pathway mediating SPP1+ macrophage-FAP+ fibroblast crosstalk, which was spatially confirmed by mIF in high-grade tumors (Gleason ≥8, p<0.001). The SPP1+FAP+ co-signature (SF-score) stratified patients into high- and low-risk groups, with high-risk patients exhibiting shorter PFS (HR=3.12, p<0.001), advanced T3-4 stage, and early biochemical recurrence, independent of PSA and Gleason score. High SF-score tumors displayed an immunosuppressive TME characterized by CTL exhaustion. Dual targeting with anti-SPP1 antibody and FAP-shRNA achieved synergistic tumor suppression and restored CD8+ T cell function.

We delineate a clinically actionable stromal-immune axis driving PCa immunosuppression. The SF-score provides a robust molecular stratification tool beyond conventional clinicopathological parameters. Therapeutic co-targeting of this axis represents a promising strategy to overcome TME-mediated therapy resistance in PCa.