Background: Prostate cancer (PCa) exhibits strong multidimensional heterogeneity. The dynamic evolution of Luminal subsets and their crosstalk with the tumor microenvironment (TME) during malignant progression remain poorly understood. Conventional single-cell transcriptomics lacks spatial context, while low-resolution spatial technologies cannot resolve single-cell and subcellular interactions. This study aimed to map a high-resolution spatial transcriptomic atlas to characterize Luminal subset heterogeneity and stromal crosstalk across progressive PCa lesions. Methods: We used high-resolution CosMx spatial molecular imaging (SMI) to profile 6 treatment-naive PCa patients and constructed a single-cell spatial transcriptomic atlas covering 385 fields of view and 154,168 high-quality cells. Unsupervised clustering, functional enrichment, cell–cell communication analysis, and flow cytometry were performed. Results: We identified 28 cell clusters, including 8 Luminal subsets (0–7) and 9 Fibroblast subsets (0–8). Pathological microregion-specific spatial distribution was observed: Luminal 0–2 dominated in BPH/PIN, while Luminal 3–7 were enriched in poorly differentiated PCa and vacuolated adenocarcinoma. These subsets were functionally linked to androgen resistance, EMT, and immune regulation. COLLAGEN, FN1, and LAMININ were identified as key pathways mediating pathological-specific tumor–stromal crosstalk. Flow cytometry verified T-cell exhaustion and immunosuppression in the TME. Conclusions: This study delineated pathological microregion-specific Luminal subset plasticity and dynamic epithelial–stromal communication across the full spectrum of PCa lesions. The uncovered molecular signatures offered mechanistic clues linking EMT, hormone resistance, and local immune status during lesion evolution. Our spatial atlas provided preliminary molecular evidence to support pathological stratification and targeted precision therapeutic strategies for PCa.

Background: Glioblastoma is highly heterogeneous, highly invasive, and resistant to temozolomide, with an extremely low 5-year survival rate. The spatial association between polyamine metabolism and ferroptosis remains unclear. Spatial metabolic heterogeneity is a core challenge in glioblastoma treatment. Methods: In this study, 17 patients with primary or recurrent high-grade glioma were included. Spatial metabolomics (n = 9), multiplex immunofluorescence (exploratory experiment n = 10, observational experiment n = 1), flow cytometry (n = 13), and mouse orthotopic models (n = 3) were combined to analyze the characteristics of the polyamine–ferroptosis axis in the para-carcinoma, tumor, and necrotic regions. Results: Differences in the polyamine-metabolism–ferroptosis axis were observed across distinct pathological regions of glioblastoma. Regional differences in polyamine distribution were associated with regional differences that lower lipid-peroxidation signals in the tumor region and higher lipid-peroxidation signals in the necrotic region. Moreover, in glioblastoma patient specimens, lipid-peroxidation-associated signals were detected in immune-cell compartments, alongside immune-cell phenotypic features suggestive of dysfunction, which show correlation with an immunosuppressive microenvironment. At the regional level, enzymes related to polyamine metabolism, namely GLS and SMS, exhibited spatial co-localization trends with lipid-peroxidation markers. Alterations in markers of the GSH-GPX4 and FSP1 antioxidant systems were spatially associated with the tumor-region-specific lipid-peroxidation profile. Partial spatial patterns of polyamine metabolism observed in human samples were partially recapitulated in mouse orthotopic tumor models. Conclusions: Glioblastoma exhibits spatially distinct distribution of the polyamine–ferroptosis-associated molecular signature. Notable correlative spatial features include spermidine and spermine enrichment, altered regional abundance within the glutamine–glutamate metabolic axis, immune-cell lipid-peroxidation-associated signals and immune-cell-dysfunction-related phenotypes, and regional shifts in GSH-GPX4/FSP1-related markers. This study identifies potential targets and provides a correlative spatial framework for region-specific therapeutic investigation, but the establishment of the causal relationship still requires further functional verification.