A scissor-guided single-cell framework defines a macrophage-derived risk score for prognostic and immunotherapy...

Macrophages are key regulators within the lung adenocarcinoma (LUAD) tumor microenvironment, and their functional states and interaction patterns are closely linked to prognosis and immunotherapy response. However, macrophage-associated risk models derived from single-cell features remain limited.

Single-cell datasets from GEO were integrated with TCGA-LUAD transcriptomic profiles to characterize macrophage subpopulations. Scissor was applied to map clinical phenotypes onto single cells and identify phenotype-associated macrophage subsets. Prognostic genes were used to construct a machine learning model termed the Scissor-Associated Macrophage Risk Score (SAMRS), which was evaluated across multiple cohorts. Genomic features, immunotherapy response indicators, cell-cell communication, virtual gene knockout, spatial transcriptomics, and functional experiments were further incorporated for mechanistic validation.

Multiple functionally distinct macrophage subsets were identified, and Scissor+ macrophages were associated with poor prognosis and oncogenic pathway activity. Communication analysis placed macrophages at central signaling positions. The SAMRS model achieved robust prognostic stratification across cohorts. High SAMRS scores were linked to higher mutation burden but stronger immune evasion signals, whereas low SAMRS scores showed more favorable predicted immunotherapy response. The model gene TIMP1 demonstrated consistent signals across bulk, single-cell, and spatial analyses, and functional experiments supported its tumor-promoting role.

A single-cell phenotype-guided macrophage risk score, SAMRS, was developed for prognostic stratification and immunotherapy response assessment in LUAD. TIMP1-related regulatory programs may contribute to macrophage-associated tumor progression.