Integrative Multi-Omics and Single-Cell Analyses Reveal that Polyamine Metabolic Reprogramming Shapes an Immunosuppressive Microenvironment and Identifies ZWINT as a Prognostic Biomarker and Therapeutic Target in Lung Adenocarcinoma

Background Polyamine metabolism has been increasingly recognized as a critical regulator of tumor progression and immune modulation; however, its cellular heterogeneity and clinical relevance in lung adenocarcinoma (LUAD) remain inadequately elucidated. Methods Single-cell RNA sequencing data from LUAD (GSE207422) were utilized to characterize the cellular distribution of polyamine metabolic activity. Polyamine-associated differentially expressed genes identified in myeloid cells were employed to construct a prognostic model using machine-learning algorithms based on TCGA datasets, with multiple GEO cohorts used for external validation. Integrative analyses, including pathway enrichment, tumor mutational profiling, immune infiltration assessment, immunotherapy response prediction, and drug sensitivity evaluation, were subsequently performed. Hub gene identification and mechanistic investigations were conducted through integrated single-cell, pseudotime trajectory, spatial transcriptomic, and transcriptional regulatory network (SCENIC-based) analyses, complemented by PCR validation in clinical samples. Results Polyamine metabolism exhibited pronounced intercellular heterogeneity and was predominantly enriched in myeloid cells. The derived prognostic model effectively stratified LUAD patients into distinct risk groups and demonstrated robust and consistent predictive performance across multiple cohorts, outperforming conventional clinical parameters and previously published signatures. High-risk patients were characterized by elevated tumor mutational burden, enhanced activation of cell cycle–related pathways, and an immunosuppressive microenvironment, indicating reduced sensitivity to immunotherapy. In contrast, low-risk patients exhibited increased immune infiltration and a higher likelihood of benefiting from immune checkpoint blockade. ZWINT was identified as a central hub gene closely associated with polyamine metabolism and adverse prognosis. Further analyses revealed that ZWINT was highly enriched in immunosuppressive macrophage subsets, particularly Macro-SPP1 and Macro-FN1, and was implicated in macrophage differentiation, intercellular communication via the SPP1–CD44 axis, and spatially elevated polyamine metabolic activity within tumor regions. Notably, SCENIC analysis uncovered a convergent transcriptional regulatory network involving E2F1, FOXM1, and MYBL2, which coordinately drive ZWINT expression, and PCR validation confirmed the pro-tumorigenic roles of these genes in LUAD. Conclusion This study provides a comprehensive and systematic characterization of polyamine metabolism in LUAD and establishes a robust prognostic framework with substantial clinical applicability. ZWINT emerges as a pivotal regulator integrating metabolic and proliferative signaling within tumor-associated macrophages, offering novel mechanistic insights and potential therapeutic targets for LUAD.