空间基因组学揭示前列腺癌中与器官特异性转移倾向相关的微环境程序

Metastatic prostate cancer (PCa) exhibits pronounced organotropism, particularly to bone, yet the spatially resolved molecular features within primary tumors that correlate with organ-specific dissemination patterns remain poorly characterized. Although bulk and single-cell transcriptomic studies have highlighted microenvironmental heterogeneity in PCa, no validated spatial genomic signatures currently exist that link intratumoral ecosystems to metastatic risk and organotropism. To address this gap, we apply high-resolution spatial genomic profiling with Xenium (10x Genomics) across human tissues and genetically engineered mouse models (GEMMs) that recapitulate PCa initiation, progression, and metastatic competence. GEMMs were used to characterize spatially resolved transcriptional programs and tumor microenvironment (TME) interactions across defined biological stages, enabling controlled dissection of the stromal, epithelial, immune, and neurovascular states associated with metastatic readiness. In parallel, we profile paired primary tumors and anatomically distinct metastatic lesions from PCa patients enrolled in the Cedars-Sinai Molecular Twin Project, including metastases to lymph nodes, bone, liver, and soft tissues. Integrative analyses demonstrate organ-associated spatial gene expression programs and multicellular niches within the primary tumor that mirror the molecular landscapes of their eventual metastatic destinations. These include bone-associated niches enriched for osteomimicry programs, ECM-remodeling fibroblasts, and osteotropic ligand-receptor signaling; nodal-associated niches defined by immune-evasive epithelial states, lymphoid-interacting stromal programs, and chemokine circuits guiding lymphatic dissemination; and visceral-associated niches characterized by metabolic rewiring, hepatotropic signaling modules, and endothelial states predisposing to vascular invasion. These spatially resolved niches highlight shared biological modules that mediate, or are associated with, organ-specific metastatic behavior. Notably, several site-specific programs are detectable within the primary tumor prior to clinical metastasis, suggesting that metastatic organotropism is encoded early through stable TME-tumor interactions. Collectively, this work establishes a cross-species framework for mapping microenvironmentally encoded metastatic trajectories and identifies spatial genomic biomarkers with potential clinical utility for predicting organ-specific metastatic risk. These findings has the potential for identifying new therapeutic strategies aimed at disrupting early metastatic niches before dissemination occurs.