Integrative multi-omics analysis defines a macrophage-associated phospholipid metabolism signature linked to immune remodeling and prognosis in lung adenocarcinoma
摘要
Phospholipid metabolism (PLM) has been implicated in macrophage polarization and tumor progression in lung adenocarcinoma (LUAD). Metabolic intermediates such as lysophosphatidylcholine modulate macrophage surface receptor signaling, while inhibition of phospholipid synthesis may attenuate tumor-associated macrophage (TAM) infiltration and suppress LUAD progression. However, macrophage-centered PLM-related prognostic signatures in LUAD remain incompletely characterized.
MethodsWe performed an integrative multi-omics analysis combining single-cell RNA sequencing, bulk transcriptomic profiling, and clinical data from TCGA and GEO cohorts. Macrophage-related genes (MacRGs) were identified using single-cell analysis and high-dimensional weighted gene co-expression network analysis (hdWGCNA). Phospholipid metabolism-related module genes were determined through differential expression analysis and WGCNA. Candidate genes were prioritized using machine learning and Cox regression analyses, followed by construction and external validation of a multigene risk model. Functional enrichment, immune microenvironment profiling, mutation landscape analysis, estimated drug sensitivity, and pseudotime trajectory analyses were performed to explore the biological relevance of the signature. Experimental validation was conducted using LUAD tissues and cell models.
ResultsA four-gene macrophage–PLM-related signature was constructed to calculate a risk score that stratified patients into high- and low-risk groups with significantly different overall survival. The risk score and T stage were independent prognostic factors, and an integrated nomogram showed acceptable prognostic performance. Functional analyses revealed distinct metabolic and signaling pathway enrichment patterns between risk groups, with 40 shared pathways across the four hub genes. The high-risk group exhibited an altered immune microenvironment characterized by differential infiltration of 20 immune cell subsets, immune checkpoint expression changes, lower ESTIMATE scores, and distinct TIDE profiles. Somatic mutation patterns and tumor mutational burden differed between groups. Drug-sensitivity estimation identified 105 compounds with differential estimated IC50 values between risk groups. DPYSL2 expression was associated with macrophage polarization dynamics, and DPYSL2 overexpression suppressed LUAD cell proliferation, migration, invasion, and clonogenicity in vitro.
ConclusionWe identified a macrophage–phospholipid metabolism-associated four-gene prognostic signature that provides prognostic information and reflects immune-metabolic remodeling in LUAD. These findings support further investigation of macrophage–phospholipid metabolism-related genes as candidate biomarkers and may inform future therapeutic stratification studies in LUAD.